from __future__ import annotations __copyright__ = """ Copyright (C) 2009-2013 Andreas Kloeckner Copyright (C) 2013- University of Illinois Board of Trustees Copyright (C) 2020 Matt Wala """ __license__ = """ Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the "Software"), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do so, subject to the following conditions: The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software. THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. """ import builtins import contextlib import dataclasses import logging import operator import re import sys from collections.abc import ( Callable, Collection, Hashable, Iterable, Iterator, Mapping, Sequence, ) from functools import reduce, wraps from sys import intern from typing import ( TYPE_CHECKING, Any, ClassVar, Concatenate, Generic, ParamSpec, Protocol, TypeVar, cast, ) from typing_extensions import Self, dataclass_transform, override from pytools.version import VERSION_TEXT if TYPE_CHECKING: from _typeshed import ReadableBuffer from typing_extensions import Self __version__ = VERSION_TEXT # These are deprecated and will go away in 2022. all = builtins.all any = builtins.any __doc__ = """ A Collection of Utilities ========================= Math ---- .. autofunction:: levi_civita Assertive accessors ------------------- .. autofunction:: one .. autofunction:: is_single_valued .. autofunction:: all_roughly_equal .. autofunction:: single_valued Memoization ----------- .. autofunction:: memoize .. autofunction:: memoize_on_first_arg .. autofunction:: memoize_method .. autofunction:: memoize_in .. autofunction:: keyed_memoize_on_first_arg .. autofunction:: keyed_memoize_method .. autofunction:: keyed_memoize_in Argmin/max ---------- .. autofunction:: argmin2 .. autofunction:: argmax2 .. autofunction:: argmin .. autofunction:: argmax Cartesian products ------------------ .. autofunction:: cartesian_product .. autofunction:: distinct_pairs Permutations, Tuples, Integer sequences --------------------------------------- .. autofunction:: wandering_element .. autofunction:: generate_nonnegative_integer_tuples_below .. autofunction:: generate_nonnegative_integer_tuples_summing_to_at_most .. autofunction:: generate_all_integer_tuples_below .. autofunction:: generate_permutations .. autofunction:: generate_unique_permutations .. autoclass:: _ConcatenableSequence Formatting ---------- .. autoclass:: Table .. autofunction:: merge_tables .. autofunction:: string_histogram .. autofunction:: word_wrap Debugging --------- .. autofunction:: typedump .. autofunction:: invoke_editor Progress bars ------------- .. autoclass:: ProgressBar Name generation --------------- .. autofunction:: generate_unique_names .. autofunction:: generate_numbered_unique_names .. autoclass:: UniqueNameGenerator Deprecation Warnings -------------------- .. autofunction:: deprecate_keyword .. autofunction:: module_getattr_for_deprecations Functions for dealing with (large) auxiliary files -------------------------------------------------- .. autofunction:: download_from_web_if_not_present Helpers for :mod:`numpy` ------------------------ .. autofunction:: reshaped_view Timing data ----------- .. autoclass:: ProcessTimer Log utilities ------------- .. autoclass:: ProcessLogger .. autoclass:: DebugProcessLogger .. autoclass:: log_process Sorting in natural order ------------------------ .. autofunction:: natorder .. autofunction:: natsorted Backports of newer Python functionality --------------------------------------- .. autofunction:: resolve_name Hashing ------- .. autoclass:: Hash .. autofunction:: unordered_hash Sampling -------- .. autofunction:: sphere_sample_equidistant .. autofunction:: sphere_sample_fibonacci String utilities ---------------- .. autofunction:: strtobool .. autofunction:: to_identifier Set-like functions for iterables -------------------------------- These functions provide set-like operations on iterables. In contrast to Python's built-in set type, they maintain the internal order of elements. .. autofunction:: unique .. autofunction:: unique_difference .. autofunction:: unique_intersection .. autofunction:: unique_union Functionality for dataclasses ----------------------------- .. autofunction:: opt_frozen_dataclass Type Variables Used ------------------- .. class:: T .. class:: R Generic unbound invariant :class:`typing.TypeVar`. .. class:: F Generic invariant :class:`typing.TypeVar` bound to a :class:`typing.Callable`. .. class:: P Generic unbound invariant :class:`typing.ParamSpec`. .. class:: _HashT """ # {{{ type variables T = TypeVar("T") R = TypeVar("R") F = TypeVar("F", bound=Callable[..., Any]) P = ParamSpec("P") K = TypeVar("K") V = TypeVar("V") # }}} # {{{ code maintenance # Undocumented on purpose for now, unclear that this is a great idea, given # that typing.deprecated exists. class MovedFunctionDeprecationWrapper: def __init__(self, f: F, deadline: int | str | None = None) -> None: if deadline is None: deadline = "the future" self.f = f self.deadline = deadline def __call__(self, *args, **kwargs): from warnings import warn warn(f"This function is deprecated and will go away in {self.deadline}. " f"Use {self.f.__module__}.{self.f.__name__} instead.", DeprecationWarning, stacklevel=2) return self.f(*args, **kwargs) def deprecate_keyword(oldkey: str, newkey: str | None = None, *, deadline: str | None = None): """Decorator used to deprecate function keyword arguments. :arg oldkey: deprecated argument name. :arg newkey: new argument name that serves the same purpose, if any. :arg deadline: expected time frame for the removal of the deprecated argument. """ from warnings import warn if deadline is None: deadline = "the future" def wrapper(func): @wraps(func) def inner_wrapper(*args, **kwargs): if oldkey in kwargs: if newkey is None: warn(f"The '{oldkey}' keyword is deprecated and will " f"go away in {deadline}.", DeprecationWarning, stacklevel=2) else: warn(f"The '{oldkey}' keyword is deprecated and will " f"go away in {deadline}. " f"Use '{newkey}' instead.", DeprecationWarning, stacklevel=2) if newkey in kwargs: raise ValueError(f"Cannot use '{oldkey}' " f"and '{newkey}' in the same call.") kwargs[newkey] = kwargs[oldkey] del kwargs[oldkey] return func(*args, **kwargs) return inner_wrapper return wrapper def module_getattr_for_deprecations( module_name: str, depr_name_to_replacement_and_obj: Mapping[ str, tuple[str, object, str | int] ], name: str ) -> object: """A helper to construct module-level :meth:`object.__getattr__` functions so that deprecated names can still be found but raise a warning. The typical usage pattern is as follows:: __getattr__ = partial(module_getattr_for_deprecations, __name__, { "OldName": ("NewName", NewName, 2026), }) """ replacement_and_obj = depr_name_to_replacement_and_obj.get(name, None) if replacement_and_obj is not None: replacement, obj, deadline = replacement_and_obj from warnings import warn warn(f"'{module_name}.{name}' is deprecated. " f"Use '{replacement}' instead. " f"'{module_name}.{name}' will continue to work until {deadline}.", DeprecationWarning, stacklevel=2) return obj raise AttributeError(name) # }}} # {{{ math def delta(x: int, y: int) -> int: if x == y: return 1 return 0 def levi_civita(tup: tuple[int, ...]) -> int: """Compute an entry of the Levi-Civita symbol for the indices *tuple*.""" if len(tup) == 2: i, j = tup return j - i if len(tup) == 3: i, j, k = tup return (j-i) * (k-i) * (k-j) // 2 raise NotImplementedError(f"Levi-Civita symbol in {len(tup)} dimensions") def norm_1(iterable): return sum(abs(x) for x in iterable) def norm_2(iterable): return sum(x**2 for x in iterable)**0.5 def norm_inf(iterable): return max(abs(x) for x in iterable) def norm_p(iterable, p): return sum(i**p for i in iterable)**(1/p) class Norm: def __init__(self, p): self.p = p def __call__(self, iterable): return sum(i**self.p for i in iterable)**(1/self.p) # }}} # {{{ data structures # {{{ record class RecordWithoutPickling: """An aggregate of named sub-variables. Assumes that each record sub-type will be individually derived from this class. """ __slots__: ClassVar[list[str]] = [] fields: ClassVar[set[str]] def __init__(self, valuedict: Mapping[str, Any] | None = None, exclude: Sequence[str] | None = None, **kwargs: Any) -> None: assert self.__class__ is not Record if exclude is None: exclude = ["self"] try: fields = self.__class__.fields except AttributeError: self.__class__.fields = fields = set() if valuedict is not None: kwargs.update(valuedict) for key, value in kwargs.items(): if key not in exclude: fields.add(key) setattr(self, key, value) def get_copy_kwargs(self, **kwargs): for f in self.__class__.fields: if f not in kwargs: with contextlib.suppress(AttributeError): kwargs[f] = getattr(self, f) return kwargs def copy(self, **kwargs) -> Self: return self.__class__(**self.get_copy_kwargs(**kwargs)) @override def __repr__(self): return "{}({})".format( self.__class__.__name__, ", ".join(f"{fld}={getattr(self, fld)!r}" for fld in sorted(self.__class__.fields) if hasattr(self, fld))) def register_fields(self, new_fields): try: fields = self.__class__.fields except AttributeError: self.__class__.fields = fields = set() fields.update(new_fields) def __getattr__(self, name): # This method is implemented to avoid pylint 'no-member' errors for # attribute access. raise AttributeError( "'{}' object has no attribute '{}'".format( self.__class__.__name__, name)) class Record(RecordWithoutPickling): __slots__: ClassVar[list[str]] = [] def __getstate__(self): return { key: getattr(self, key) for key in self.__class__.fields if hasattr(self, key)} def __setstate__(self, valuedict): try: fields = self.__class__.fields except AttributeError: self.__class__.fields = fields = set() for key, value in valuedict.items(): fields.add(key) setattr(self, key, value) @override def __eq__(self, other): if self is other: return True return (self.__class__ == other.__class__ and self.__getstate__() == other.__getstate__()) @override def __ne__(self, other): return not self.__eq__(other) class ImmutableRecordWithoutPickling(RecordWithoutPickling): """Hashable record. Does not explicitly enforce immutability.""" _cached_hash: int | None def __init__(self, *args, **kwargs): RecordWithoutPickling.__init__(self, *args, **kwargs) self._cached_hash = None @override def __hash__(self) -> int: # This attribute may vanish during pickling. if getattr(self, "_cached_hash", None) is None: self._cached_hash = hash(( type(self), *(getattr(self, field) for field in self.__class__.fields) )) assert self._cached_hash is not None return self._cached_hash class ImmutableRecord(ImmutableRecordWithoutPickling, Record): pass # }}} class Reference: def __init__(self, value): self.value = value def get(self): from warnings import warn warn("Reference.get() is deprecated -- use ref.value instead. " "This will stop working in 2025.", stacklevel=2) return self.value def set(self, value): self.value = value class FakeList: def __init__(self, f, length): self._Length = length self._Function = f def __len__(self): return self._Length def __getitem__(self, index): try: return [self._Function(i) for i in range(*index.indices(self._Length))] except AttributeError: return self._Function(index) # {{{ dependent dictionary class DependentDictionary: def __init__(self, f, start=None): if start is None: start = {} self._Function = f self._Dictionary = start.copy() def copy(self): return DependentDictionary(self._Function, self._Dictionary) def __contains__(self, key): try: self[key] return True except KeyError: return False def __getitem__(self, key): try: return self._Dictionary[key] except KeyError: return self._Function(self._Dictionary, key) def __setitem__(self, key, value): self._Dictionary[key] = value def genuineKeys(self): # noqa: N802 return list(self._Dictionary.keys()) def iteritems(self): return self._Dictionary.items() def iterkeys(self): return self._Dictionary.keys() def itervalues(self): return self._Dictionary.values() # }}} # }}} # {{{ assertive accessors def one(iterable: Iterable[T]) -> T: """Return the first entry of *iterable*. Assert that *iterable* has only that one entry. """ it = iter(iterable) try: v = next(it) except StopIteration: raise ValueError("empty iterable passed to 'one()'") from None def no_more(): try: next(it) raise ValueError("iterable with more than one entry passed to 'one()'") except StopIteration: return True assert no_more() return v def is_single_valued( iterable: Iterable[T], equality_pred: Callable[[T, T], bool] = operator.eq ) -> bool: it = iter(iterable) try: first_item = next(it) except StopIteration: raise ValueError("empty iterable passed to 'single_valued()'") from None return all(equality_pred(other_item, first_item) for other_item in it) all_equal = is_single_valued def all_roughly_equal(iterable, threshold): return is_single_valued(iterable, equality_pred=lambda a, b: abs(a-b) < threshold) def single_valued( iterable: Iterable[T], equality_pred: Callable[[T, T], bool] = operator.eq ) -> T: """Return the first entry of *iterable*; Assert that other entries are the same with the first entry of *iterable*. """ it = iter(iterable) try: first_item = next(it) except StopIteration: raise ValueError("empty iterable passed to 'single_valued()'") from None assert all(equality_pred(other_item, first_item) for other_item in it) return first_item # }}} # {{{ memoization / attribute storage def memoize(*args: F, **kwargs: Any) -> F: """Stores previously computed function values in a cache. Two keyword-only arguments are supported: :arg use_kwargs: Allows the caller to use keyword arguments. Defaults to ``False``. Setting this to ``True`` has a non-negligible performance impact. :arg key: A function receiving the same arguments as the decorated function which computes and returns the cache key. """ use_kw = bool(kwargs.pop("use_kwargs", False)) default_key_func: Callable[..., Any] | None if use_kw: def default_key_func(*inner_args, **inner_kwargs): return inner_args, frozenset(inner_kwargs.items()) else: default_key_func = None key_func = kwargs.pop("key", default_key_func) if kwargs: raise TypeError( "memoize received unexpected keyword arguments: {}".format( ", ".join(kwargs.keys()))) if key_func is not None: def _decorator(func): def wrapper(*args, **kwargs): key = key_func(*args, **kwargs) try: return func._memoize_dic[key] except AttributeError: # _memoize_dic doesn't exist yet. result = func(*args, **kwargs) func._memoize_dic = {key: result} return result except KeyError: result = func(*args, **kwargs) func._memoize_dic[key] = result return result from functools import update_wrapper update_wrapper(wrapper, func) return wrapper else: def _decorator(func): def wrapper(*args): try: return func._memoize_dic[args] except AttributeError: # _memoize_dic doesn't exist yet. result = func(*args) func._memoize_dic = {args: result} return result except KeyError: result = func(*args) func._memoize_dic[args] = result return result from functools import update_wrapper update_wrapper(wrapper, func) return wrapper if not args: return _decorator # type: ignore[return-value] if callable(args[0]) and len(args) == 1: return _decorator(args[0]) raise TypeError( f"memoize received unexpected position arguments: {args}") FunctionValueCache = memoize class _HasKwargs: pass def memoize_on_first_arg( function: Callable[Concatenate[T, P], R], *, cache_dict_name: str | None = None) -> Callable[Concatenate[T, P], R]: """Like :func:`memoize_method`, but for functions that take the object in which do memoization information is stored as first argument. Supports cache deletion via ``function_name.clear_cache(self)``. """ if cache_dict_name is None: cache_dict_name = intern( f"_memoize_dic_{function.__module__}{function.__name__}" ) def wrapper(obj: T, *args: P.args, **kwargs: P.kwargs) -> R: key = (_HasKwargs, frozenset(kwargs.items()), *args) if kwargs else args assert cache_dict_name is not None try: return getattr(obj, cache_dict_name)[key] except AttributeError: attribute_error = True except KeyError: attribute_error = False result = function(obj, *args, **kwargs) if attribute_error: object.__setattr__(obj, cache_dict_name, {key: result}) return result getattr(obj, cache_dict_name)[key] = result return result def clear_cache(obj): object.__delattr__(obj, cache_dict_name) from functools import update_wrapper new_wrapper = update_wrapper(wrapper, function) # type-ignore because mypy has a point here, stuffing random attributes # into the function's dict is moderately sketchy. new_wrapper.clear_cache = clear_cache # type: ignore[attr-defined] return new_wrapper def memoize_method( method: Callable[Concatenate[T, P], R] ) -> Callable[Concatenate[T, P], R]: """Supports cache deletion via ``method_name.clear_cache(self)``. .. versionchanged:: 2021.2 Can memoize methods on classes that do not allow setting attributes (e.g. by overwriting ``__setattr__``), e.g. frozen :mod:`dataclasses`. """ return memoize_on_first_arg(method, cache_dict_name=intern(f"_memoize_dic_{method.__name__}")) class keyed_memoize_on_first_arg(Generic[T, P, R]): # noqa: N801 """Like :func:`memoize_method`, but for functions that take the object in which memoization information is stored as first argument. Supports cache deletion via ``function_name.clear_cache(self)``. :arg key: A function receiving the same arguments as the decorated function which computes and returns the cache key. :arg cache_dict_name: The name of the `dict` attribute in the instance used to hold the cache. .. versionadded :: 2020.3 """ def __init__(self, key: Callable[P, Hashable], *, cache_dict_name: str | None = None) -> None: self.key = key self.cache_dict_name = cache_dict_name def _default_cache_dict_name(self, function: Callable[Concatenate[T, P], R]) -> str: return intern(f"_memoize_dic_{function.__module__}{function.__name__}") def __call__( self, function: Callable[Concatenate[T, P], R] ) -> Callable[Concatenate[T, P], R]: cache_dict_name = self.cache_dict_name key = self.key if cache_dict_name is None: cache_dict_name = self._default_cache_dict_name(function) def wrapper(obj: T, *args: P.args, **kwargs: P.kwargs) -> R: cache_key = key(*args, **kwargs) assert cache_dict_name is not None try: return getattr(obj, cache_dict_name)[cache_key] except AttributeError: result = function(obj, *args, **kwargs) object.__setattr__(obj, cache_dict_name, {cache_key: result}) return result except KeyError: result = function(obj, *args, **kwargs) getattr(obj, cache_dict_name)[cache_key] = result return result def clear_cache(obj): object.__delattr__(obj, cache_dict_name) from functools import update_wrapper new_wrapper = update_wrapper(wrapper, function) new_wrapper.clear_cache = clear_cache # type: ignore[attr-defined] return new_wrapper class keyed_memoize_method(keyed_memoize_on_first_arg): # noqa: N801 """Like :class:`memoize_method`, but additionally uses a function *key* to compute the key under which the function result is stored. Supports cache deletion via ``method_name.clear_cache(self)``. :arg key: A function receiving the same arguments as the decorated function which computes and returns the cache key. .. versionadded :: 2020.3 .. versionchanged:: 2021.2 Can memoize methods on classes that do not allow setting attributes (e.g. by overwriting ``__setattr__``), e.g. frozen :mod:`dataclasses`. """ @override def _default_cache_dict_name(self, function): return intern(f"_memoize_dic_{function.__name__}") class memoize_in: # noqa: N801 """Adds a cache to the function it decorates. The cache is attached to *container* and must be uniquely specified by *identifier* (i.e. all functions using the same *container* and *identifier* will be using the same cache). The decorated function may only receive positional arguments. .. note:: This function works well on nested functions, which do not have stable global identifiers. .. versionchanged :: 2020.3 *identifier* no longer needs to be a :class:`str`, but it needs to be hashable. .. versionchanged:: 2021.2.1 Can now use instances of classes as *container* that do not allow setting attributes (e.g. by overwriting ``__setattr__``), e.g. frozen :mod:`dataclasses`. """ def __init__(self, container: Any, identifier: Hashable) -> None: try: memoize_in_dict = container._pytools_memoize_in_dict except AttributeError: memoize_in_dict = {} object.__setattr__(container, "_pytools_memoize_in_dict", memoize_in_dict) self.cache_dict = memoize_in_dict.setdefault(identifier, {}) def __call__(self, inner: Callable[P, R]) -> Callable[P, R]: @wraps(inner) def new_inner(*args: P.args, **kwargs: P.kwargs) -> R: assert not kwargs try: return self.cache_dict[args] except KeyError: result = inner(*args, **kwargs) self.cache_dict[args] = result return result return new_inner class keyed_memoize_in(Generic[P]): # noqa: N801 """Like :class:`memoize_in`, but additionally uses a function *key* to compute the key under which the function result is memoized. :arg key: A function receiving the same arguments as the decorated function which computes and returns the cache key. .. versionadded :: 2021.2.1 """ def __init__(self, container: Any, identifier: Hashable, key: Callable[P, Hashable]) -> None: try: memoize_in_dict = container._pytools_keyed_memoize_in_dict except AttributeError: memoize_in_dict = {} object.__setattr__(container, "_pytools_keyed_memoize_in_dict", memoize_in_dict) self.cache_dict = memoize_in_dict.setdefault(identifier, {}) self.key = key def __call__(self, inner: Callable[P, R]) -> Callable[P, R]: @wraps(inner) def new_inner(*args: P.args, **kwargs: P.kwargs) -> R: assert not kwargs key = self.key(*args, **kwargs) try: return self.cache_dict[key] except KeyError: result = inner(*args, **kwargs) self.cache_dict[key] = result return result return new_inner # }}} # {{{ syntactical sugar class InfixOperator: """Pseudo-infix operators that allow syntax of the kind `op1 <> op2'. Following a recipe from http://aspn.activestate.com/ASPN/Cookbook/Python/Recipe/384122 """ def __init__(self, function): self.function = function def __rlshift__(self, other): return InfixOperator(lambda x: self.function(other, x)) def __rshift__(self, other): return self.function(other) def call(self, a, b): return self.function(a, b) def monkeypatch_method(cls): # from GvR, http://mail.python.org/pipermail/python-dev/2008-January/076194.html def decorator(func): setattr(cls, func.__name__, func) return func return decorator def monkeypatch_class(_name, bases, namespace): # from GvR, http://mail.python.org/pipermail/python-dev/2008-January/076194.html assert len(bases) == 1, "Exactly one base class required" base = bases[0] for name, value in namespace.items(): if name != "__metaclass__": setattr(base, name, value) return base # }}} # {{{ generic utilities def add_tuples(t1, t2): return tuple(t1v + t2v for t1v, t2v in zip(t1, t2, strict=True)) def negate_tuple(t1): return tuple(-t1v for t1v in t1) def shift(vec, dist): """Return a copy of *vec* shifted by *dist* such that .. code:: python shift(a, i)[j] == a[(i+j) % len(a)] """ result = vec[:] N = len(vec) # noqa: N806 dist = dist % N # modulo only returns positive distances! if dist > 0: result[dist:] = vec[:N-dist] result[:dist] = vec[N-dist:] return result def len_iterable(iterable): return sum(1 for i in iterable) def flatten(iterable): """For an iterable of sub-iterables, generate each member of each sub-iterable in turn, i.e. a flattened version of that super-iterable. Example: Turn [[a,b,c],[d,e,f]] into [a,b,c,d,e,f]. """ for sublist in iterable: yield from sublist def general_sum(sequence): return reduce(operator.add, sequence) def linear_combination(coefficients, vectors): result = coefficients[0] * vectors[0] for c, v in zip(coefficients[1:], vectors[1:], strict=True): result += c*v return result def common_prefix(iterable, empty=None): it = iter(iterable) try: pfx = next(it) except StopIteration: return empty for v in it: for j, pfx_j in enumerate(pfx): if pfx_j != v[j]: pfx = pfx[:j] if j == 0: return pfx break return pfx def decorate(function, iterable): return [(x, function(x)) for x in iterable] def partition(criterion, iterable): part_true = [] part_false = [] for i in iterable: if criterion(i): part_true.append(i) else: part_false.append(i) return part_true, part_false def partition2(iterable): part_true = [] part_false = [] for pred, i in iterable: if pred: part_true.append(i) else: part_false.append(i) return part_true, part_false def product(iterable: Iterable[T]) -> T | int: from operator import mul return reduce(mul, iterable, 1) def reverse_dictionary(the_dict: Mapping[K, V]) -> dict[V, K]: result: dict[V, K] = {} for key, value in the_dict.items(): if value in result: raise RuntimeError( f"non-reversible mapping, duplicate key '{value}'") result[value] = key return result def set_sum(set_iterable): from operator import or_ return reduce(or_, set_iterable, set()) def div_ceil(nr, dr): return -(-nr // dr) def uniform_interval_splitting(n, granularity, max_intervals): """ Return *(interval_size, num_intervals)* such that:: num_intervals * interval_size >= n and:: (num_intervals - 1) * interval_size < n and *interval_size* is a multiple of *granularity*. """ # ported from Thrust grains = div_ceil(n, granularity) # one grain per interval if grains <= max_intervals: return granularity, grains grains_per_interval = div_ceil(grains, max_intervals) interval_size = grains_per_interval * granularity num_intervals = div_ceil(n, interval_size) return interval_size, num_intervals def find_max_where(predicate, prec=1e-5, initial_guess=1, fail_bound=1e38): """Find the largest value for which a predicate is true, along a half-line. 0 is assumed to be the lower bound.""" # {{{ establish bracket mag = initial_guess if predicate(mag): mag *= 2 while predicate(mag): mag *= 2 if mag > fail_bound: raise RuntimeError("predicate appears to be true " f"everywhere, up to {fail_bound:g}") lower_true = mag/2 upper_false = mag else: mag /= 2 while not predicate(mag): mag /= 2 if mag < prec: return mag lower_true = mag upper_false = mag*2 # }}} # {{{ refine # Refine a bracket between *lower_true*, where the predicate is true, # and *upper_false*, where it is false, until *prec* is satisfied. assert predicate(lower_true) assert not predicate(upper_false) while abs(lower_true-upper_false) > prec: mid = (lower_true+upper_false)/2 if predicate(mid): lower_true = mid else: upper_false = mid return lower_true # }}} # }}} # {{{ argmin, argmax def argmin2(iterable, return_value=False): it = iter(iterable) try: current_argmin, current_min = next(it) except StopIteration: raise ValueError("argmin of empty iterable") from None for arg, item in it: if item < current_min: current_argmin = arg current_min = item if return_value: return current_argmin, current_min return current_argmin def argmax2(iterable, return_value=False): it = iter(iterable) try: current_argmax, current_max = next(it) except StopIteration: raise ValueError("argmax of empty iterable") from None for arg, item in it: if item > current_max: current_argmax = arg current_max = item if return_value: return current_argmax, current_max return current_argmax def argmin(iterable): return argmin2(enumerate(iterable)) def argmax(iterable): return argmax2(enumerate(iterable)) # }}} # {{{ cartesian products etc. def cartesian_product(*args): if len(args) == 1: for arg in args[0]: yield (arg,) return first = args[:-1] for prod in cartesian_product(*first): for i in args[-1]: yield (*prod, i) def distinct_pairs(list1, list2): for i, xi in enumerate(list1): for j, yj in enumerate(list2): if i != j: yield (xi, yj) def cartesian_product_sum(list1, list2): """This routine returns a list of sums of each element of list1 with each element of list2. Also works with lists. """ for i in list1: for j in list2: yield i+j # }}} # {{{ elementary statistics def average(iterable): """Return the average of the values in iterable. iterable may not be empty. """ it = iterable.__iter__() try: s = next(it) count = 1 except StopIteration: raise ValueError("empty average") from None for value in it: s = s + value count += 1 return s/count class VarianceAggregator: """Online variance calculator. See http://en.wikipedia.org/wiki/Algorithms_for_calculating_variance Adheres to pysqlite's aggregate interface. """ def __init__(self, entire_pop): self.n = 0 self.mean = 0 self.m2 = 0 self.entire_pop = entire_pop def step(self, x): self.n += 1 delta_ = x - self.mean self.mean += delta_/self.n self.m2 += delta_*(x - self.mean) def finalize(self): if self.entire_pop: if self.n == 0: return None return self.m2/self.n if self.n <= 1: return None return self.m2/(self.n - 1) def variance(iterable, entire_pop): v_comp = VarianceAggregator(entire_pop) for x in iterable: v_comp.step(x) return v_comp.finalize() def std_deviation(iterable, finite_pop): from math import sqrt return sqrt(variance(iterable, finite_pop)) # }}} # {{{ permutations, tuples, integer sequences def wandering_element(length: int, wanderer: int = 1, landscape: int = 0) -> Iterator[tuple[int, ...]]: for i in range(length): yield i*(landscape,) + (wanderer,) + (length-1-i)*(landscape,) def indices_in_shape(shape): from warnings import warn warn("indices_in_shape is deprecated. You should prefer numpy.ndindex.", DeprecationWarning, stacklevel=2) if isinstance(shape, int): shape = (shape,) if not shape: yield () elif len(shape) == 1: for i in range(shape[0]): yield (i,) else: remainder = shape[1:] for i in range(shape[0]): for rest in indices_in_shape(remainder): yield (i, *rest) def generate_nonnegative_integer_tuples_below( n: Sequence[int] | int, length: int | None = None, least: int = 0 ) -> Iterator[tuple[int, ...]]: """n may be a sequence, in which case length must be None.""" if length is None: assert not isinstance(n, int) if not n: yield () return my_n = n[0] n = n[1:] next_length = None else: assert isinstance(n, int) my_n = n assert length >= 0 if length == 0: yield () return next_length = length-1 for i in range(least, my_n): my_part = (i,) for base in generate_nonnegative_integer_tuples_below(n, next_length, least): yield my_part + base def generate_decreasing_nonnegative_tuples_summing_to( n: int, length: int, min_value: int = 0, max_value: int | None = None ) -> Iterator[tuple[int, ...]]: if length == 0: yield () elif length == 1: if max_value is None or n <= max_value: yield (n,) else: return else: if max_value is None or n < max_value: max_value = n for i in range(min_value, max_value + 1): for remainder in generate_decreasing_nonnegative_tuples_summing_to( n - i, length - 1, min_value=min_value, max_value=i): yield (i, *remainder) def generate_nonnegative_integer_tuples_summing_to_at_most( n: int, length: int) -> Iterator[tuple[int, ...]]: """Enumerate all non-negative integer tuples summing to at most n, exhausting the search space by varying the first entry fastest, and the last entry the slowest. """ assert length >= 0 if length == 0: yield () else: for i in range(n+1): for remainder in generate_nonnegative_integer_tuples_summing_to_at_most( n-i, length-1): yield (*remainder, i) # backwards compatibility generate_positive_integer_tuples_below = generate_nonnegative_integer_tuples_below def _pos_and_neg_adaptor( tuple_iter: Iterator[tuple[int, ...]] ) -> Iterator[tuple[int, ...]]: for tup in tuple_iter: nonzero_indices = [i for i in range(len(tup)) if tup[i] != 0] for do_neg_tup in generate_nonnegative_integer_tuples_below( 2, len(nonzero_indices)): this_result = list(tup) for index, do_neg in enumerate(do_neg_tup): if do_neg: this_result[nonzero_indices[index]] *= -1 yield tuple(this_result) def generate_all_integer_tuples_below( n: int, length: int, least_abs: int = 0 ) -> Iterator[tuple[int, ...]]: return _pos_and_neg_adaptor(generate_nonnegative_integer_tuples_below( n, length, least_abs)) T_co = TypeVar("T_co", covariant=True) class _ConcatenableSequence(Generic[T_co], Protocol): """ A protocol that supports the following: .. automethod:: __getitem__ .. automethod:: __add__ .. automethod:: __len__ """ def __getitem__(self, slice: slice, /) -> Self: ... def __add__(self, value: Self, /) -> Self: ... def __len__(self) -> int: ... def __iter__(self) -> Iterator[T_co]: ... def generate_permutations( original: _ConcatenableSequence[T] ) -> Iterator[_ConcatenableSequence[T]]: """Generate all permutations of the list *original*. Nicked from http://aspn.activestate.com/ASPN/Cookbook/Python/Recipe/252178 """ if len(original) <= 1: yield original else: for perm_ in generate_permutations(original[1:]): for i in range(len(perm_)+1): # NOTE: ary[0:1] works in both string and list contexts yield perm_[:i] + original[0:1] + perm_[i:] def generate_unique_permutations( original: _ConcatenableSequence[T] ) -> Iterator[_ConcatenableSequence[T]]: """Generate all unique permutations of the list *original*. Note that, unlike for :func:`generate_permutations`, *original* must be a hashable object. """ had_those = set() for perm_ in generate_permutations(original): if perm_ not in had_those: had_those.add(perm_) yield perm_ def enumerate_basic_directions(dimensions): coordinate_list = [[0], [1], [-1]] return reduce(cartesian_product_sum, [coordinate_list] * dimensions)[1:] # }}} # {{{ graph algorithms from pytools.graph import a_star as a_star_moved a_star = MovedFunctionDeprecationWrapper(a_star_moved) # }}} # {{{ formatting # {{{ table formatting class Table: """An ASCII table generator. .. automethod:: __init__ .. automethod:: add_row .. autoproperty:: nrows .. autoproperty:: ncolumns .. automethod:: __str__ .. automethod:: github_markdown .. automethod:: csv .. automethod:: latex .. automethod:: text_without_markup """ def __init__(self, alignments: tuple[str, ...] | None = None) -> None: """Create a new :class:`Table`. :arg alignments: A :class:`tuple` of alignments of each column: ``"l"``, ``"c"``, or ``"r"``, for left, center, and right alignment, respectively). Columns which have no alignment specifier will use the last specified alignment. For example, with ``alignments=("l", "r")``, the third and all following columns will use right alignment. """ if alignments is None: alignments = ("l",) else: if any(a not in ("l", "c", "r") for a in alignments): raise ValueError(f"alignments are ('l', 'c', 'r'): {alignments}") alignments = tuple(alignments) self.rows: list[tuple[str, ...]] = [] self.alignments = alignments @property def nrows(self) -> int: """The number of rows currently in the table.""" return len(self.rows) @property def ncolumns(self) -> int: """The number of columns currently in the table.""" return len(self.rows[0]) def add_row(self, row: tuple[Any, ...]) -> None: """Add *row* to the table. Note that all rows must have the same number of columns.""" if self.rows and len(row) != self.ncolumns: raise ValueError( f"tried to add a row with {len(row)} columns to " f"a table with {self.ncolumns} columns") self.rows.append(tuple(str(i) for i in row)) def _get_alignments(self) -> tuple[str, ...]: # NOTE: If not all alignments were specified, extend alignments with the # last alignment specified return ( self.alignments + (self.alignments[-1],) * (self.ncolumns - len(self.alignments)) )[:self.ncolumns] def _get_column_widths(self, rows) -> tuple[int, ...]: return tuple( max(len(row[i]) for row in rows) for i in range(self.ncolumns) ) @override def __str__(self) -> str: """ Returns a string representation of the table. .. doctest :: >>> tbl = Table(alignments=['l', 'r', 'l']) >>> tbl.add_row([1, '|']) >>> tbl.add_row([10, '20||']) >>> print(tbl) 1 | | ---+------ 10 | 20|| """ if not self.rows: return "" alignments = self._get_alignments() col_widths = self._get_column_widths(self.rows) lines = [" | ".join([ cell.center(cwidth) if align == "c" else cell.ljust(cwidth) if align == "l" else cell.rjust(cwidth) for cell, cwidth, align in zip(row, col_widths, alignments, strict=True)]) for row in self.rows] lines[1:1] = ["+".join("-" * (cwidth + 1 + (i > 0)) for i, cwidth in enumerate(col_widths))] return "\n".join(lines) def str_with_maxlen(self, maxlen: int) -> str: """ Returns a string representation of the table, limiting lines to *maxlen*. Only makes sense with 'left' alignment in the last column. .. doctest :: >>> tbl = Table(alignments=['l', 'l']) >>> tbl.add_row([0, 'normal column']) >>> tbl.add_row([1, 'very very very very long column']) >>> tbl.add_row([10, '20']) >>> print(tbl.str_with_maxlen(15)) 0 | normal ... ---+----------- 1 | very v ... 10 | 20 """ if self._get_alignments()[-1] != "l": from warnings import warn warn("str_with_maxlen() only makes sense with 'left' " "alignment in the last column", stacklevel=2) lines = str(self).split("\n") lines = [line.rstrip() for line in lines] for i, line in enumerate(lines): if len(line) > maxlen: if i == 1: # Header separator line lines[i] = line[:maxlen] else: lines[i] = line[:maxlen-4] + " ..." return "\n".join(lines) def github_markdown(self) -> str: r"""Returns a string representation of the table formatted as `GitHub-Flavored Markdown. `__ .. doctest :: >>> tbl = Table(alignments=['l', 'r', 'l']) >>> tbl.add_row([1, '|']) >>> tbl.add_row([10, '20||']) >>> print(tbl.github_markdown()) 1 | \| :--|-------: 10 | 20\|\| """ if not self.rows: return "" def escape(cell: str) -> str: # Pipe symbols ('|') must be replaced return cell.replace("|", "\\|") rows = [tuple(escape(cell) for cell in row) for row in self.rows] alignments = self._get_alignments() col_widths = self._get_column_widths(rows) lines = [" | ".join([ cell.center(cwidth) if align == "c" else cell.ljust(cwidth) if align == "l" else cell.rjust(cwidth) for cell, cwidth, align in zip(row, col_widths, alignments, strict=True)]) for row in rows] lines[1:1] = ["|".join( (":" + "-" * (cwidth - 1 + (i > 0)) + ":") if align == "c" else (":" + "-" * (cwidth + (i > 0))) if align == "l" else ("-" * (cwidth + (i > 0)) + ":") for i, (cwidth, align) in enumerate( zip(col_widths, alignments, strict=True)))] return "\n".join(lines) def csv(self, dialect: str = "excel", csv_kwargs: dict[str, Any] | None = None) -> str: """Returns a string containing a CSV representation of the table. :arg dialect: String passed to :func:`csv.writer`. :arg csv_kwargs: Dict of arguments passed to :func:`csv.writer`. .. doctest :: >>> tbl = Table() >>> tbl.add_row([1, ","]) >>> tbl.add_row([10, 20]) >>> print(tbl.csv()) 1,"," 10,20 """ if not self.rows: return "" import csv import io if csv_kwargs is None: csv_kwargs = {} # Default is "\r\n" if "lineterminator" not in csv_kwargs: csv_kwargs["lineterminator"] = "\n" output = io.StringIO() writer = csv.writer(output, dialect, **csv_kwargs) writer.writerows(self.rows) return output.getvalue().rstrip(csv_kwargs["lineterminator"]) def latex(self, skip_lines: int = 0, hline_after: tuple[int, ...] | None = None) -> str: r"""Returns a string containing the rows of a LaTeX representation of the table. :arg skip_lines: number of lines to skip at the start of the table. :arg hline_after: list of row indices after which to add an ``hline`` (the indices must subtract *skip_lines*, if non-zero). .. doctest:: >>> tbl = Table() >>> tbl.add_row([0, "skipped"]) >>> tbl.add_row([1, "apple"]) >>> tbl.add_row([2, "pear"]) >>> print(tbl.latex(skip_lines=1)) 1 & apple \\ 2 & pear \\ """ if not self.rows: return "" if hline_after is None: hline_after = () lines = [] for row_nr, row in enumerate(self.rows[skip_lines:]): lines.append(fr"{' & '.join(row)} \\") if row_nr in hline_after: lines.append(r"\hline") return "\n".join(lines) def text_without_markup(self) -> str: """Returns a string representation of the table without markup. .. doctest:: >>> tbl = Table() >>> tbl.add_row([0, "orange"]) >>> tbl.add_row([1111, "apple"]) >>> tbl.add_row([2, "pear"]) >>> print(tbl.text_without_markup()) 0 orange 1111 apple 2 pear """ if not self.rows: return "" alignments = self._get_alignments() col_widths = self._get_column_widths(self.rows) lines = [" ".join([ cell.center(cwidth) if align == "c" else cell.ljust(cwidth) if align == "l" else cell.rjust(cwidth) for cell, cwidth, align in zip(row, col_widths, alignments, strict=True)]) for row in self.rows] # Remove the extra space added by the last cell lines = [line.rstrip() for line in lines] return "\n".join(lines) def merge_tables(*tables: Table, skip_columns: tuple[int, ...] | None = None) -> Table: """ :arg skip_columns: a :class:`tuple` of column indices to skip in all the tables except the first one. """ if len(tables) == 1: return tables[0] if any(tables[0].nrows != tbl.nrows for tbl in tables[1:]): raise ValueError("tables do not have the same number of rows") if isinstance(skip_columns, int): skip_columns = (skip_columns,) def remove_columns(i, row): if i == 0 or skip_columns is None: return row return tuple( entry for i, entry in enumerate(row) if i not in skip_columns ) alignments = sum(( remove_columns(i, tbl._get_alignments()) for i, tbl in enumerate(tables) ), ()) result = Table(alignments=alignments) for i in range(tables[0].nrows): row = [] for j, tbl in enumerate(tables): row.extend(remove_columns(j, tbl.rows[i])) result.add_row(tuple(row)) return result # }}} # {{{ histogram formatting def string_histogram( iterable, min_value=None, max_value=None, bin_count=20, width=70, bin_starts=None, use_unicode=True): if bin_starts is None: if min_value is None or max_value is None: iterable = list(iterable) min_value = min(iterable) max_value = max(iterable) bin_width = (max_value - min_value)/bin_count bin_starts = [min_value+bin_width*i for i in range(bin_count)] bins = [0 for i in range(len(bin_starts))] from bisect import bisect for value in iterable: if (max_value is not None and value > max_value) or value < bin_starts[0]: from warnings import warn warn("string_histogram: out-of-bounds value ignored", stacklevel=2) else: bin_nr = bisect(bin_starts, value)-1 try: bins[bin_nr] += 1 except Exception: print(value, bin_nr, bin_starts) raise from math import ceil, floor if use_unicode: def format_bar(cnt): scaled = cnt*width/max_count full = floor(scaled) eighths = ceil((scaled-full)*8) if eighths: return full*chr(0x2588) + chr(0x2588+(8-eighths)) return full*chr(0x2588) else: def format_bar(cnt): return ceil(cnt*width/max_count)*"#" max_count = max(bins) total_count = sum(bins) return "\n".join("{:9g} |{:9d} | {:3.0f} % | {}".format( bin_start, bin_value, bin_value/total_count*100, format_bar(bin_value)) for bin_start, bin_value in zip(bin_starts, bins, strict=True)) # }}} def word_wrap(text, width, wrap_using="\n"): # http://code.activestate.com/recipes/148061-one-liner-word-wrap-function/ r""" A word-wrap function that preserves existing line breaks and most spaces in the text. Expects that existing line breaks are posix newlines (``\n``). """ space_or_break = [" ", wrap_using] return reduce(lambda line, word: "{}{}{}".format( line, space_or_break[( len(line) - line.rfind("\n") - 1 + len(word.split("\n", 1)[0]) ) >= width], word), text.split(" ") ) # }}} # {{{ debugging class StderrToStdout: def __enter__(self): self.stderr_backup = sys.stderr sys.stderr = sys.stdout def __exit__(self, exc_type, exc_val, exc_tb): sys.stderr = self.stderr_backup del self.stderr_backup def typedump(val: Any, max_seq: int = 5, special_handlers: Mapping[type, Callable] | None = None, fully_qualified_name: bool = True) -> str: """ Return a string representation of the type of *val*, recursing into iterable objects. :arg val: The object for which the type should be returned. :arg max_seq: For iterable objects, the maximum number of elements to include in the return string. Lower this value if you get a :class:`RecursionError`. :arg special_handlers: An optional mapping of specific types to special handlers. :arg fully_qualified_name: Return fully qualified names, that is, include module names and use ``__qualname__`` instead of ``__name__``. :returns: A string representation of the type of *val*. """ if special_handlers is None: special_handlers = {} try: hdlr = special_handlers[type(val)] except KeyError: pass else: return hdlr(val) def objname(obj: Any) -> str: if type(obj).__module__ == "builtins": if fully_qualified_name: return type(obj).__qualname__ return type(obj).__name__ if fully_qualified_name: return type(obj).__module__ + "." + type(obj).__qualname__ return type(obj).__name__ # Special handling for 'str' since it is also iterable if isinstance(val, str): return "str" try: len(val) except TypeError: return objname(val) else: if isinstance(val, dict): return "{%s}" % ( ", ".join( f"{str(k)!r}: {typedump(v)}" for k, v in val.items())) try: if len(val) > max_seq: t = ",".join(typedump(x, max_seq, special_handlers) for x in val[:max_seq]) return f"{objname(val)}({t},...)" t = ",".join(typedump(x, max_seq, special_handlers) for x in val) return f"{objname(val)}({t})" except TypeError: return objname(val) def invoke_editor(s, filename="edit.txt", descr="the file"): from tempfile import mkdtemp tempdir = mkdtemp() from os.path import join full_name = join(tempdir, filename) with open(full_name, "w") as outf: outf.write(str(s)) import os if "EDITOR" in os.environ: from subprocess import Popen p = Popen([os.environ["EDITOR"], full_name]) os.waitpid(p.pid, 0) else: print("(Set the EDITOR environment variable to be " "dropped directly into an editor next time.)") input(f"Edit {descr} at {full_name} now, then hit [Enter]:") with open(full_name) as inf: result = inf.read() return result # }}} # {{{ progress bars class ProgressBar: """ .. automethod:: draw .. automethod:: progress .. automethod:: set_progress .. automethod:: finished .. automethod:: __enter__ .. automethod:: __exit__ """ def __init__(self, descr: str, total: int, initial: int = 0, length: int = 40) -> None: import time self.description = descr self.total = total self.done = initial self.length = length self.last_squares = -1 self.start_time = time.time() self.last_update_time = self.start_time self.speed_meas_start_time = self.start_time self.speed_meas_start_done = initial self.time_per_step: float | None = None def draw(self) -> None: import time now = time.time() squares = int(self.done/self.total*self.length) if squares != self.last_squares or now-self.last_update_time > 0.5: if (self.done != self.speed_meas_start_done and now-self.speed_meas_start_time > 3): new_time_per_step = (now-self.speed_meas_start_time) \ / (self.done-self.speed_meas_start_done) if self.time_per_step is not None: self.time_per_step = (new_time_per_step + self.time_per_step)/2 else: self.time_per_step = new_time_per_step self.speed_meas_start_time = now self.speed_meas_start_done = self.done if self.time_per_step is not None: eta_str = "{:7.1f}s ".format( max(0, (self.total-self.done) * self.time_per_step) ) else: eta_str = "?" sys.stderr.write("{:<20} [{}] ETA {}\r".format( self.description, squares*"#"+(self.length-squares)*" ", eta_str)) self.last_squares = squares self.last_update_time = now def progress(self, steps: int = 1) -> None: self.set_progress(self.done + steps) def set_progress(self, done: int) -> None: self.done = done self.draw() def finished(self) -> None: self.set_progress(self.total) sys.stderr.write("\n") def __enter__(self) -> None: self.draw() def __exit__(self, exc_type: Any, exc_val: Any, exc_tb: Any) -> None: self.finished() # }}} # {{{ file system related def assert_not_a_file(name): import os if os.access(name, os.F_OK): raise OSError(f"file `{name}' already exists") def add_python_path_relative_to_script(rel_path): from os.path import abspath, dirname, join script_name = sys.argv[0] rel_script_dir = dirname(script_name) sys.path.append(abspath(join(rel_script_dir, rel_path))) # }}} # {{{ numpy dtype mangling def common_dtype(dtypes, default=None): dtypes = list(dtypes) if dtypes: return argmax2((dtype, dtype.num) for dtype in dtypes) if default is not None: return default raise ValueError( "cannot find common dtype of empty dtype list") def to_uncomplex_dtype(dtype): import numpy as np return np.array(1, dtype=dtype).real.dtype.type def match_precision(dtype, dtype_to_match): import numpy tgt_is_double = dtype_to_match in [ numpy.float64, numpy.complex128] dtype_is_complex = dtype.kind == "c" if dtype_is_complex: if tgt_is_double: return numpy.dtype(numpy.complex128) return numpy.dtype(numpy.complex64) if tgt_is_double: return numpy.dtype(numpy.float64) return numpy.dtype(numpy.float32) # }}} # {{{ unique name generation def generate_unique_names(prefix): yield prefix try_num = 0 while True: yield f"{prefix}_{try_num}" try_num += 1 UNIQUE_NAME_GEN_COUNTER_RE = re.compile(r"^(?P\w+)_(?P\d+)$") def generate_numbered_unique_names( prefix: str, num: int | None = None, suffix: str = "") -> Iterable[tuple[int, str]]: if num is None: yield (0, prefix + suffix) num = 0 while True: name = f"{prefix}_{num}{suffix}" num += 1 yield (num, name) generate_unique_possibilities = MovedFunctionDeprecationWrapper( generate_unique_names) class UniqueNameGenerator: """ Class that creates a new :class:`str` on each :meth:`__call__` that is unique to the generator. .. automethod:: __init__ .. automethod:: is_name_conflicting .. automethod:: add_name .. automethod:: add_names .. automethod:: __call__ """ def __init__(self, existing_names: Collection[str] | None = None, forced_prefix: str = "", forced_suffix: str = "") -> None: """ Create a new :class:`UniqueNameGenerator`. :arg existing_names: a :class:`set` of existing names that will be skipped when generating new names. :arg forced_prefix: all generated :class:`str` have this prefix. :arg forced_suffix: all generated :class:`str` have this suffix. """ if existing_names is None: existing_names = set() self.existing_names = set(existing_names) self.forced_prefix = forced_prefix self.forced_suffix: str = forced_suffix self.prefix_to_counter: dict[str, int] = {} def is_name_conflicting(self, name: str) -> bool: """Returns *True* if *name* conflicts with an existing :class:`str`.""" return name in self.existing_names def _name_added(self, name: str) -> None: """Callback to alert subclasses when a name has been added. .. note:: This will not get called for the names in the *existing_names* argument to :meth:`__init__`. """ def add_name(self, name: str, *, conflicting_ok: bool = False) -> None: """ :arg conflicting_ok: A flag to dictate the behavior when *name* is conflicting with the set of existing names. If *True*, a conflict is silently passed. If *False*, a :class:`ValueError` is raised on encountering a conflict. """ if (not conflicting_ok) and self.is_name_conflicting(name): raise ValueError(f"name '{name}' conflicts with existing names") if not name.startswith(self.forced_prefix): raise ValueError( f"name '{name}' does not start with required prefix " f"'{self.forced_prefix}'") self.existing_names.add(name) self._name_added(name) def add_names(self, names: Iterable[str], *, conflicting_ok: bool = False) -> None: """ :arg conflicting_ok: Plainly passed to :meth:`UniqueNameGenerator.add_name`. """ for name in names: self.add_name(name, conflicting_ok=conflicting_ok) def __call__(self, based_on: str = "id") -> str: """Returns a new unique name.""" based_on = self.forced_prefix + based_on counter = self.prefix_to_counter.get(based_on, None) # {{{ try to get counter from based_on if not already present if counter is None: counter_match = UNIQUE_NAME_GEN_COUNTER_RE.match(based_on) if counter_match: based_on = counter_match.groupdict()["based_on"] counter = int(counter_match.groupdict()["counter"]) # }}} for counter, var_name in generate_numbered_unique_names( # noqa: B020,B007 based_on, counter, self.forced_suffix): if not self.is_name_conflicting(var_name): break self.prefix_to_counter[based_on] = counter var_name = intern(var_name) # pylint: disable=undefined-loop-variable self.existing_names.add(var_name) self._name_added(var_name) return var_name # }}} # {{{ recursion limit class MinRecursionLimit: def __init__(self, min_rec_limit): self.min_rec_limit = min_rec_limit def __enter__(self): self.prev_recursion_limit = sys.getrecursionlimit() new_limit = max(self.prev_recursion_limit, self.min_rec_limit) sys.setrecursionlimit(new_limit) def __exit__(self, exc_type, exc_val, exc_tb): # Deep recursion can produce deeply nested data structures # (or long chains of to-be gc'd generators) that cannot be # undergo garbage collection with a lower recursion limit. # # As a result, it doesn't seem possible to lower the recursion limit # again after it has been raised without causing reliability issues. # # See https://gitlab.tiker.net/inducer/sumpy/issues/31 for # context. pass # }}} # {{{ download from web if not present def download_from_web_if_not_present(url, local_name=None): """ .. versionadded:: 2017.5 """ from os.path import basename, exists if local_name is None: local_name = basename(url) if not exists(local_name): from urllib.request import Request, urlopen from pytools.version import VERSION_TEXT req = Request(url, headers={ "User-Agent": f"pytools/{VERSION_TEXT}" }) with urlopen(req) as inf: contents = inf.read() with open(local_name, "wb") as outf: outf.write(contents) # }}} # {{{ find git revisions def find_git_revision(tree_root): # Keep this routine self-contained so that it can be copy-pasted into # setup.py. from os.path import abspath, exists, join tree_root = abspath(tree_root) if not exists(join(tree_root, ".git")): return None # construct minimal environment # stolen from # https://github.com/numpy/numpy/blob/055ce3e90b50b5f9ef8cf1b8641c42e391f10735/setup.py#L70-L92 import os env = {} for k in ["SYSTEMROOT", "PATH", "HOME"]: v = os.environ.get(k) if v is not None: env[k] = v # LANGUAGE is used on win32 env["LANGUAGE"] = "C" env["LANG"] = "C" env["LC_ALL"] = "C" from subprocess import PIPE, STDOUT, Popen p = Popen(["git", "rev-parse", "HEAD"], shell=False, stdin=PIPE, stdout=PIPE, stderr=STDOUT, close_fds=True, cwd=tree_root, env=env) (git_rev, _) = p.communicate() git_rev = git_rev.decode() git_rev = git_rev.rstrip() retcode = p.returncode assert retcode is not None if retcode != 0: from warnings import warn warn("unable to find git revision", stacklevel=1) return None return git_rev def find_module_git_revision(module_file, n_levels_up): from os.path import dirname, join tree_root = join(*([dirname(module_file), ".." * n_levels_up])) return find_git_revision(tree_root) # }}} # {{{ create a reshaped view of a numpy array def reshaped_view(a, newshape): """ Create a new view object with shape ``newshape`` without copying the data of ``a``. This function is different from ``numpy.reshape`` by raising an exception when data copy is necessary. :arg a: a :class:`numpy.ndarray` object. :arg newshape: an ``int`` object or a tuple of ``int`` objects. .. versionadded:: 2018.4 """ newview = a.view() newview.shape = newshape return newview # }}} # {{{ process timer SUPPORTS_PROCESS_TIME = True class ProcessTimer: """Measures elapsed wall time and process time. .. automethod:: __enter__ .. automethod:: __exit__ .. automethod:: done Timing data attributes: .. attribute:: wall_elapsed .. attribute:: process_elapsed .. versionadded:: 2018.5 """ def __init__(self): import time self.perf_counter_start = time.perf_counter() self.process_time_start = time.process_time() self.wall_elapsed = None self.process_elapsed = None def __enter__(self): return self def __exit__(self, exc_type, exc_val, exc_tb): self.done() def done(self): import time self.wall_elapsed = time.perf_counter() - self.perf_counter_start self.process_elapsed = time.process_time() - self.process_time_start @override def __str__(self): cpu = self.process_elapsed / self.wall_elapsed return f"{self.wall_elapsed:.2f}s wall {cpu:.2f}x CPU" @override def __repr__(self): wall = self.wall_elapsed process = self.process_elapsed return (f"{type(self).__name__}" f"(wall_elapsed={wall!r}s, process_elapsed={process!r}s)") # }}} # {{{ log utilities def _log_start_if_long(logger, sleep_duration, done_indicator, noisy_level, description): from time import sleep sleep(sleep_duration) if not done_indicator[0]: logger.log( noisy_level, "%s: started %.gs ago", description, sleep_duration) class ProcessLogger: """Logs the completion time of a (presumably) lengthy process to :mod:`logging`. Only uses a high log level if the process took perceptible time. .. automethod:: __init__ .. automethod:: done .. automethod:: __enter__ .. automethod:: __exit__ """ default_noisy_level = logging.INFO def __init__( self, logger, description, silent_level=None, noisy_level=None, long_threshold_seconds=None): self.logger = logger self.description = description self.silent_level = silent_level or logging.DEBUG self.noisy_level = noisy_level or self.default_noisy_level self.long_threshold_seconds = ( # 0 is a valid value that should override the default 0.3 if long_threshold_seconds is None else long_threshold_seconds) self.logger.log(self.silent_level, "%s: start", self.description) self._done_indicator = [False] import threading self.late_start_log_thread = threading.Thread( target=_log_start_if_long, args=(logger, 10*self.long_threshold_seconds, self._done_indicator, self.noisy_level, self.description), # Do not delay interpreter exit if thread not finished. daemon=True) # https://github.com/firedrakeproject/firedrake/issues/1422 # Starting a thread may irrecoverably break various environments, # e.g. MPI. # # Since the late-start logging is an optional 'quality-of-life' # feature for interactive use, do not do it unless there is (weak) # evidence of interactive use. import sys if sys.stdin is None: # Can happen, e.g., if pudb is controlling the console. use_late_start_logging = False elif hasattr(sys.stdin, "closed") and not sys.stdin.closed: # can query stdin.isatty() only if stdin's open use_late_start_logging = sys.stdin.isatty() else: use_late_start_logging = False import os if os.environ.get("PYTOOLS_LOG_NO_THREADS", ""): use_late_start_logging = False if use_late_start_logging: # https://github.com/firedrakeproject/firedrake/issues/1422 # # Starting a thread may fail in various environments, e.g. MPI. # Since the late-start logging is an optional 'quality-of-life' # feature for interactive use, tolerate failures of it without # warning. with contextlib.suppress(RuntimeError): self.late_start_log_thread.start() self.timer = ProcessTimer() def done( self, extra_msg=None, *extra_fmt_args): self.timer.done() self._done_indicator[0] = True completion_level = ( self.noisy_level if self.timer.wall_elapsed > self.long_threshold_seconds else self.silent_level) msg = "%s: completed (%s)" fmt_args = [self.description, str(self.timer)] if extra_msg: msg = f"{msg}: {extra_msg}" fmt_args.extend(extra_fmt_args) self.logger.log(completion_level, msg, *fmt_args) def __enter__(self): pass def __exit__(self, exc_type, exc_val, exc_tb): self.done() class DebugProcessLogger(ProcessLogger): default_noisy_level = logging.DEBUG class log_process: # noqa: N801 """A decorator that uses :class:`ProcessLogger` to log data about calls to the wrapped function. .. automethod:: __init__ .. automethod:: __call__ """ def __init__(self, logger, description=None, long_threshold_seconds=None): self.logger = logger self.description = description self.long_threshold_seconds = long_threshold_seconds def __call__(self, wrapped): def wrapper(*args, **kwargs): with ProcessLogger( self.logger, self.description or wrapped.__name__, long_threshold_seconds=self.long_threshold_seconds): return wrapped(*args, **kwargs) from functools import update_wrapper new_wrapper = update_wrapper(wrapper, wrapped) return new_wrapper # }}} # {{{ sorting in natural order def natorder(item): """Return a key for natural order string comparison. See :func:`natsorted`. .. versionadded:: 2020.1 """ import re result = [] for (int_val, string_val) in re.findall(r"(\d+)|(\D+)", item): if int_val: result.append(int(int_val)) # Tie-breaker in case of leading zeros in *int_val*. Longer values # compare smaller to preserve order of numbers in decimal notation, # e.g., "1.001" < "1.01" # (cf. https://github.com/sourcefrog/natsort) result.append(-len(int_val)) else: result.append(string_val) return result def natsorted(iterable, key=None, reverse=False): """Sort using natural order [1]_, as opposed to lexicographic order. Example:: >>> sorted(["_10", "_1", "_9"]) == ["_1", "_10", "_9"] True >>> natsorted(["_10", "_1", "_9"]) == ["_1", "_9", "_10"] True :arg iterable: an iterable to be sorted. It must only have strings, unless *key* is specified. :arg key: if provided, a key function that returns strings for ordering using natural order. :arg reverse: if *True*, sorts in descending order. :returns: a sorted list .. [1] https://en.wikipedia.org/wiki/Natural_sort_order .. versionadded:: 2020.1 """ if key is None: def key(x): return x return sorted(iterable, key=lambda y: natorder(key(y)), reverse=reverse) # }}} # {{{ resolve_name # https://github.com/python/cpython/commit/1ed61617a4a6632905ad6a0b440cd2cafb8b6414 _DOTTED_WORDS = r"[a-z_]\w*(\.[a-z_]\w*)*" _NAME_PATTERN = re.compile(f"^({_DOTTED_WORDS})(:({_DOTTED_WORDS})?)?$", re.I) del _DOTTED_WORDS def resolve_name(name): """A backport of :func:`pkgutil.resolve_name` (added in Python 3.9). .. versionadded:: 2021.1.2 """ from warnings import warn warn("'pytools.resolve_name' is deprecated and will be removed in 2024. " "Use 'pkgutil.resolve_name' from the standard library instead.", DeprecationWarning, stacklevel=2) import pkgutil return pkgutil.resolve_name(name) # }}} # {{{ unordered_hash class Hash(Protocol): """A protocol for the hashes from :mod:`hashlib`. .. automethod:: update .. automethod:: digest .. automethod:: hexdigest .. automethod:: copy """ def update(self, obj: ReadableBuffer, /) -> None: ... def digest(self) -> bytes: ... def hexdigest(self) -> str: ... def copy(self) -> Self: ... @property def digest_size(self) -> int: ... @property def name(self) -> str: ... _HashT = TypeVar("_HashT", bound=Hash) def unordered_hash(hash_instance: _HashT, iterable: Iterable[ReadableBuffer], hash_constructor: Callable[[], _HashT] | None = None) -> _HashT: """Using a hash algorithm given by the parameter-less constructor *hash_constructor*, return a hash object whose internal state depends on the entries of *iterable*, but not their order. If *hash* is the instance returned by evaluating ``hash_constructor()``, then the each entry *i* of the iterable must permit ``hash.update(i)`` to succeed. An example of *hash_constructor* is ``hashlib.sha256`` from :mod:`hashlib`. ``hash.digest_size`` must also be defined. If *hash_constructor* is not provided, ``hash_instance.name`` is used to deduce it. :returns: the updated *hash_instance*. .. warning:: The construction used in this function is likely not cryptographically secure. Do not use this function in a security-relevant context. .. versionadded:: 2021.2 """ if hash_constructor is None: import hashlib from functools import partial hash_constructor = cast( "Callable[[], _HashT]", partial(hashlib.new, hash_instance.name)) assert hash_constructor is not None h_int = 0 for i in iterable: h_i = hash_constructor() h_i.update(i) # Using sys.byteorder (for efficiency) here technically makes the # hash system-dependent (which it should not be), however the # effect of this is undone by the to_bytes conversion below, while # left invariant by the intervening XOR operations (which do not # mix adjacent bits). h_int = h_int ^ int.from_bytes(h_i.digest(), sys.byteorder) hash_instance.update(h_int.to_bytes(hash_instance.digest_size, sys.byteorder)) return hash_instance # }}} # {{{ sphere_sample def sphere_sample_equidistant(npoints_approx: int, r: float = 1.0): """Generate points regularly distributed on a sphere based on https://www.cmu.edu/biolphys/deserno/pdf/sphere_equi.pdf. :returns: an :class:`~numpy.ndarray` of shape ``(3, npoints)``, where ``npoints`` does not generally equal *npoints_approx*. """ import numpy as np points: list[np.ndarray] = [] count = 0 a = 4 * np.pi / npoints_approx d = a ** 0.5 M_theta = int(np.ceil(np.pi / d)) # noqa: N806 d_theta = np.pi / M_theta d_phi = a / d_theta for m in range(M_theta): theta = np.pi * (m + 0.5) / M_theta M_phi = int(np.ceil(2 * np.pi * np.sin(theta) / d_phi)) # noqa: N806 for n in range(M_phi): phi = 2 * np.pi * n / M_phi points.append(np.array([ r * np.sin(theta) * np.cos(phi), r * np.sin(theta) * np.sin(phi), r * np.cos(theta) ])) count += 1 # add poles for i in range(3): for sign in [-1, +1]: pole = np.zeros(3) pole[i] = r * sign points.append(pole) return np.array(points).T.copy() # NOTE: each tuple contains ``(epsilon, max_npoints)`` _SPHERE_FIBONACCI_OFFSET = ( (0.33, 24), (1.33, 177), (3.33, 890), (10, 11000), (27, 39000), (75, 600000), (214, float("inf")), ) def sphere_sample_fibonacci( npoints: int, r: float = 1.0, *, optimize: str | None = None): """Generate points on a sphere based on an offset Fibonacci lattice from [2]_. .. [2] http://extremelearning.com.au/how-to-evenly-distribute-points-on-a-sphere-more-effectively-than-the-canonical-fibonacci-lattice/ :param optimize: takes the values: *None* to use the standard Fibonacci lattice, ``"minimum"`` to minimize the nearest neighbor distances in the lattice and ``"average"`` to minimize the average distances in the lattice. :returns: an :class:`~numpy.ndarray` of shape ``(3, npoints)``. """ import numpy as np if optimize is None: epsilon = 0.5 elif optimize == "minimum": epsilon, _ = next(o for o in _SPHERE_FIBONACCI_OFFSET if npoints < o[1]) elif optimize == "average": epsilon = 0.36 else: raise ValueError(f"unknown 'optimize' choice: '{optimize}'") golden_ratio = (1 + np.sqrt(5)) / 2 n = np.arange(npoints) phi = 2.0 * np.pi * n / golden_ratio theta = np.arccos(1.0 - 2.0 * (n + epsilon) / (npoints + 2 * epsilon - 1)) return np.stack([ r * np.sin(theta) * np.cos(phi), r * np.sin(theta) * np.sin(phi), r * np.cos(theta) ]) # }}} # {{{ strtobool def strtobool(val: str | None, default: bool | None = None) -> bool: """Convert a string representation of truth to True or False. True values are 'y', 'yes', 't', 'true', 'on', and '1'; false values are 'n', 'no', 'f', 'false', 'off', and '0'. Uppercase versions are also accepted. If *default* is None, raises ValueError if *val* is anything else. If *val* is None and *default* is not None, returns *default*. Based on :func:`distutils.util.strtobool`. :param val: Value to convert. :param default: Value to return if *val* is None. :returns: Truth value of *val*. """ if val is None and default is not None: return default if val is None: raise ValueError(f"invalid truth value '{val}'. " "Valid values are ('y', 'yes', 't', 'true', 'on', '1') " "for 'True' and ('n', 'no', 'f', 'false', 'off', '0') " "for 'False'. Uppercase versions are also accepted.") val = val.lower() if val in ("y", "yes", "t", "true", "on", "1"): return True if val in ("n", "no", "f", "false", "off", "0"): return False raise ValueError(f"invalid truth value '{val}'. " "Valid values are ('y', 'yes', 't', 'true', 'on', '1') " "for 'True' and ('n', 'no', 'f', 'false', 'off', '0') " "for 'False'. Uppercase versions are also accepted.") # }}} # {{{ to_identifier def to_identifier(s: str) -> str: """Convert a string to a valid Python identifier, by removing non-alphanumeric, non-underscore characters, and prepending an underscore if the string starts with a numeric character. :param s: The string to convert to an identifier. :returns: The converted string. """ if s.isidentifier(): return s s = "".join(c for c in s if c.isalnum() or c == "_") if len(s) == 0: return "_" if s[0].isdigit(): s = "_" + s return s # }}} # {{{ unique def unique(seq: Iterable[T]) -> Collection[T]: """Return unique elements in *seq*, removing all duplicates. The internal order of the elements is preserved. See also :func:`itertools.groupby` (which removes consecutive duplicates).""" return dict.fromkeys(seq) def unique_difference(*args: Iterable[T]) -> Collection[T]: r"""Return unique elements that are in the first iterable in *\*args* but not in any of the others. The internal order of the elements is preserved.""" if not args: return [] res = dict.fromkeys(args[0]) for seq in args[1:]: for item in seq: if item in res: del res[item] return res def unique_intersection(*args: Iterable[T]) -> Collection[T]: r"""Return unique elements that are common to all iterables in *\*args*. The internal order of the elements is preserved.""" if not args: return [] res = dict.fromkeys(args[0]) for seq in args[1:]: seq = set(seq) res = {item: None for item in res if item in seq} return res def unique_union(*args: Iterable[T]) -> Collection[T]: r"""Return unique elements that are in any iterable in *\*args*. The internal order of the elements is preserved.""" if not args: return [] res: dict[T, None] = {} for seq in args: for item in seq: if item not in res: res[item] = None return res # }}} @dataclass_transform(frozen_default=True, field_specifiers=(dataclasses.field,)) def opt_frozen_dataclass( *, init: bool = True, repr: bool = True, eq: bool = True, order: bool = False, unsafe_hash: bool | None = None, match_args: bool = True, kw_only: bool = False, slots: bool = False, # Added in 3.11. weakref_slot: bool = False, ) -> Callable[[type[T]], type[T]]: """Like :func:`dataclasses.dataclass`, but marks the dataclass frozen only if :data:`__debug__` is active. Frozen dataclasses have a ~20% cost penalty (on creation, from having to call :meth:`object.__setattr__`) that this decorator avoids when the interpreter runs with "optimization" enabled. The resulting dataclass supports hashing, even when it is not actually frozen, if *unsafe_hash* is left at the default or set to *True*. .. note:: Python prevents non-frozen dataclasses from inheriting from frozen ones, and vice versa. To ensure frozen-ness is applied predictably in all scenarios (mainly :data:`__debug__` on and off), it is strongly recommended that all dataclasses inheriting from ones with this decorator *also* use this decorator. There are no run-time checks to make sure of this. .. versionadded:: 2024.1.18 """ def map_cls(cls: type[T]) -> type[T]: # This ensures that the resulting dataclass is hashable with and without # __debug__, unless the user overrides unsafe_hash or provides their own # __hash__ method. if unsafe_hash is None: if (eq and not __debug__ and "__hash__" not in cls.__dict__): loc_unsafe_hash = True else: loc_unsafe_hash = False else: loc_unsafe_hash = unsafe_hash dc_extra_kwargs: dict[str, bool] = {} if weakref_slot: if sys.version_info < (3, 11): raise TypeError("weakref_slot is not available before Python 3.11") dc_extra_kwargs["weakref_slot"] = weakref_slot from dataclasses import dataclass return dataclass( init=init, repr=repr, eq=eq, order=order, unsafe_hash=loc_unsafe_hash, frozen=__debug__, match_args=match_args, kw_only=kw_only, slots=slots, **dc_extra_kwargs, )(cls) return map_cls def _test(): import doctest doctest.testmod() if __name__ == "__main__": _test() # vim: foldmethod=marker