import copy import os import pathlib import re import string import sys import types import warnings from contextlib import contextmanager from enum import Enum from textwrap import dedent from typing import ( Any, Dict, Iterator, List, Optional, Tuple, Type, Union, get_type_hints, ) import yaml from .errors import ( ConfigIndexError, ConfigTypeError, ConfigValueError, GrammarParseError, OmegaConfBaseException, ValidationError, ) from .grammar_parser import SIMPLE_INTERPOLATION_PATTERN, parse try: import dataclasses except ImportError: # pragma: no cover dataclasses = None # type: ignore # pragma: no cover try: import attr except ImportError: # pragma: no cover attr = None # type: ignore # pragma: no cover NoneType: Type[None] = type(None) BUILTIN_VALUE_TYPES: Tuple[Type[Any], ...] = ( int, float, bool, str, bytes, NoneType, ) # Regexprs to match key paths like: a.b, a[b], ..a[c].d, etc. # We begin by matching the head (in these examples: a, a, ..a). # This can be read as "dots followed by any character but `.` or `[`" # Note that a key starting with brackets, like [a], is purposedly *not* # matched here and will instead be handled in the next regex below (this # is to keep this regex simple). KEY_PATH_HEAD = re.compile(r"(\.)*[^.[]*") # Then we match other keys. The following expression matches one key and can # be read as a choice between two syntaxes: # - `.` followed by anything except `.` or `[` (ex: .b, .d) # - `[` followed by anything then `]` (ex: [b], [c]) KEY_PATH_OTHER = re.compile(r"\.([^.[]*)|\[(.*?)\]") # source: https://yaml.org/type/bool.html YAML_BOOL_TYPES = [ "y", "Y", "yes", "Yes", "YES", "n", "N", "no", "No", "NO", "true", "True", "TRUE", "false", "False", "FALSE", "on", "On", "ON", "off", "Off", "OFF", ] class Marker: def __init__(self, desc: str): self.desc = desc def __repr__(self) -> str: return self.desc # To be used as default value when `None` is not an option. _DEFAULT_MARKER_: Any = Marker("_DEFAULT_MARKER_") class OmegaConfDumper(yaml.Dumper): # type: ignore str_representer_added = False @staticmethod def str_representer(dumper: yaml.Dumper, data: str) -> yaml.ScalarNode: with_quotes = yaml_is_bool(data) or is_int(data) or is_float(data) return dumper.represent_scalar( yaml.resolver.BaseResolver.DEFAULT_SCALAR_TAG, data, style=("'" if with_quotes else None), ) def get_omega_conf_dumper() -> Type[OmegaConfDumper]: if not OmegaConfDumper.str_representer_added: OmegaConfDumper.add_representer(str, OmegaConfDumper.str_representer) OmegaConfDumper.str_representer_added = True return OmegaConfDumper def yaml_is_bool(b: str) -> bool: return b in YAML_BOOL_TYPES def get_yaml_loader() -> Any: class OmegaConfLoader(yaml.SafeLoader): # type: ignore def construct_mapping(self, node: yaml.Node, deep: bool = False) -> Any: keys = set() for key_node, value_node in node.value: if key_node.tag != yaml.resolver.BaseResolver.DEFAULT_SCALAR_TAG: continue if key_node.value in keys: raise yaml.constructor.ConstructorError( "while constructing a mapping", node.start_mark, f"found duplicate key {key_node.value}", key_node.start_mark, ) keys.add(key_node.value) return super().construct_mapping(node, deep=deep) loader = OmegaConfLoader loader.add_implicit_resolver( "tag:yaml.org,2002:float", re.compile( """^(?: [-+]?[0-9]+(?:_[0-9]+)*\\.[0-9_]*(?:[eE][-+]?[0-9]+)? |[-+]?[0-9]+(?:_[0-9]+)*(?:[eE][-+]?[0-9]+) |\\.[0-9]+(?:_[0-9]+)*(?:[eE][-+][0-9]+)? |[-+]?[0-9]+(?:_[0-9]+)*(?::[0-5]?[0-9])+\\.[0-9_]* |[-+]?\\.(?:inf|Inf|INF) |\\.(?:nan|NaN|NAN))$""", re.X, ), list("-+0123456789."), ) loader.yaml_implicit_resolvers = { key: [ (tag, regexp) for tag, regexp in resolvers if tag != "tag:yaml.org,2002:timestamp" ] for key, resolvers in loader.yaml_implicit_resolvers.items() } loader.add_constructor( "tag:yaml.org,2002:python/object/apply:pathlib.Path", lambda loader, node: pathlib.Path(*loader.construct_sequence(node)), ) loader.add_constructor( "tag:yaml.org,2002:python/object/apply:pathlib.PosixPath", lambda loader, node: pathlib.PosixPath(*loader.construct_sequence(node)), ) loader.add_constructor( "tag:yaml.org,2002:python/object/apply:pathlib.WindowsPath", lambda loader, node: pathlib.WindowsPath(*loader.construct_sequence(node)), ) return loader def _get_class(path: str) -> type: from importlib import import_module module_path, _, class_name = path.rpartition(".") mod = import_module(module_path) try: klass: type = getattr(mod, class_name) except AttributeError: raise ImportError(f"Class {class_name} is not in module {module_path}") return klass def is_union_annotation(type_: Any) -> bool: if sys.version_info >= (3, 10): # pragma: no cover if isinstance(type_, types.UnionType): return True return getattr(type_, "__origin__", None) is Union def _resolve_optional(type_: Any) -> Tuple[bool, Any]: """Check whether `type_` is equivalent to `typing.Optional[T]` for some T.""" if is_union_annotation(type_): args = type_.__args__ if NoneType in args: optional = True args = tuple(a for a in args if a is not NoneType) else: optional = False if len(args) == 1: return optional, args[0] elif len(args) >= 2: return optional, Union[args] else: assert False if type_ is Any: return True, Any if type_ in (None, NoneType): return True, NoneType return False, type_ def _is_optional(obj: Any, key: Optional[Union[int, str]] = None) -> bool: """Check `obj` metadata to see if the given node is optional.""" from .base import Container, Node if key is not None: assert isinstance(obj, Container) obj = obj._get_node(key) assert isinstance(obj, Node) return obj._is_optional() def _resolve_forward(type_: Type[Any], module: str) -> Type[Any]: import typing # lgtm [py/import-and-import-from] forward = typing.ForwardRef if hasattr(typing, "ForwardRef") else typing._ForwardRef # type: ignore if type(type_) is forward: return _get_class(f"{module}.{type_.__forward_arg__}") else: if is_dict_annotation(type_): kt, vt = get_dict_key_value_types(type_) if kt is not None: kt = _resolve_forward(kt, module=module) if vt is not None: vt = _resolve_forward(vt, module=module) return Dict[kt, vt] # type: ignore if is_list_annotation(type_): et = get_list_element_type(type_) if et is not None: et = _resolve_forward(et, module=module) return List[et] # type: ignore if is_tuple_annotation(type_): its = get_tuple_item_types(type_) its = tuple(_resolve_forward(it, module=module) for it in its) return Tuple[its] # type: ignore return type_ def extract_dict_subclass_data(obj: Any, parent: Any) -> Optional[Dict[str, Any]]: """Check if obj is an instance of a subclass of Dict. If so, extract the Dict keys/values.""" from omegaconf.omegaconf import _maybe_wrap is_type = isinstance(obj, type) obj_type = obj if is_type else type(obj) subclasses_dict = is_dict_subclass(obj_type) if subclasses_dict: warnings.warn( f"Class `{obj_type.__name__}` subclasses `Dict`." + " Subclassing `Dict` in Structured Config classes is deprecated," + " see github.com/omry/omegaconf/issues/663", UserWarning, stacklevel=9, ) if is_type: return None elif subclasses_dict: dict_subclass_data = {} key_type, element_type = get_dict_key_value_types(obj_type) for name, value in obj.items(): is_optional, type_ = _resolve_optional(element_type) type_ = _resolve_forward(type_, obj.__module__) try: dict_subclass_data[name] = _maybe_wrap( ref_type=type_, is_optional=is_optional, key=name, value=value, parent=parent, ) except ValidationError as ex: format_and_raise( node=None, key=name, value=value, cause=ex, msg=str(ex) ) return dict_subclass_data else: return None def get_attr_class_fields(obj: Any) -> List["attr.Attribute[Any]"]: is_type = isinstance(obj, type) obj_type = obj if is_type else type(obj) fields = attr.fields_dict(obj_type).values() return [f for f in fields if f.metadata.get("omegaconf_ignore") is not True] def get_attr_data(obj: Any, allow_objects: Optional[bool] = None) -> Dict[str, Any]: from omegaconf.omegaconf import OmegaConf, _maybe_wrap flags = {"allow_objects": allow_objects} if allow_objects is not None else {} from omegaconf import MISSING d = {} is_type = isinstance(obj, type) obj_type = obj if is_type else type(obj) dummy_parent = OmegaConf.create({}, flags=flags) dummy_parent._metadata.object_type = obj_type resolved_hints = get_type_hints(obj_type) for attrib in get_attr_class_fields(obj): name = attrib.name is_optional, type_ = _resolve_optional(resolved_hints[name]) type_ = _resolve_forward(type_, obj.__module__) if not is_type: value = getattr(obj, name) else: value = attrib.default if value == attr.NOTHING: value = MISSING if is_union_annotation(type_) and not is_supported_union_annotation(type_): e = ConfigValueError( f"Unions of containers are not supported:\n{name}: {type_str(type_)}" ) format_and_raise(node=None, key=None, value=value, cause=e, msg=str(e)) try: d[name] = _maybe_wrap( ref_type=type_, is_optional=is_optional, key=name, value=value, parent=dummy_parent, ) except (ValidationError, GrammarParseError) as ex: format_and_raise( node=dummy_parent, key=name, value=value, cause=ex, msg=str(ex) ) d[name]._set_parent(None) dict_subclass_data = extract_dict_subclass_data(obj=obj, parent=dummy_parent) if dict_subclass_data is not None: d.update(dict_subclass_data) return d def get_dataclass_fields(obj: Any) -> List["dataclasses.Field[Any]"]: fields = dataclasses.fields(obj) return [f for f in fields if f.metadata.get("omegaconf_ignore") is not True] def get_dataclass_data( obj: Any, allow_objects: Optional[bool] = None ) -> Dict[str, Any]: from omegaconf.omegaconf import MISSING, OmegaConf, _maybe_wrap flags = {"allow_objects": allow_objects} if allow_objects is not None else {} d = {} is_type = isinstance(obj, type) obj_type = get_type_of(obj) dummy_parent = OmegaConf.create({}, flags=flags) dummy_parent._metadata.object_type = obj_type resolved_hints = get_type_hints(obj_type) for field in get_dataclass_fields(obj): name = field.name is_optional, type_ = _resolve_optional(resolved_hints[field.name]) type_ = _resolve_forward(type_, obj.__module__) has_default = field.default != dataclasses.MISSING has_default_factory = field.default_factory != dataclasses.MISSING if not is_type: value = getattr(obj, name) else: if has_default: value = field.default elif has_default_factory: value = field.default_factory() # type: ignore else: value = MISSING if is_union_annotation(type_) and not is_supported_union_annotation(type_): e = ConfigValueError( f"Unions of containers are not supported:\n{name}: {type_str(type_)}" ) format_and_raise(node=None, key=None, value=value, cause=e, msg=str(e)) try: d[name] = _maybe_wrap( ref_type=type_, is_optional=is_optional, key=name, value=value, parent=dummy_parent, ) except (ValidationError, GrammarParseError) as ex: format_and_raise( node=dummy_parent, key=name, value=value, cause=ex, msg=str(ex) ) d[name]._set_parent(None) dict_subclass_data = extract_dict_subclass_data(obj=obj, parent=dummy_parent) if dict_subclass_data is not None: d.update(dict_subclass_data) return d def is_dataclass(obj: Any) -> bool: from omegaconf.base import Node if dataclasses is None or isinstance(obj, Node): return False return dataclasses.is_dataclass(obj) def is_attr_class(obj: Any) -> bool: from omegaconf.base import Node if attr is None or isinstance(obj, Node): return False return attr.has(obj) def is_structured_config(obj: Any) -> bool: return is_attr_class(obj) or is_dataclass(obj) def is_dataclass_frozen(type_: Any) -> bool: return type_.__dataclass_params__.frozen # type: ignore def is_attr_frozen(type_: type) -> bool: # This is very hacky and probably fragile as well. # Unfortunately currently there isn't an official API in attr that can detect that. # noinspection PyProtectedMember return type_.__setattr__ == attr._make._frozen_setattrs # type: ignore def get_type_of(class_or_object: Any) -> Type[Any]: type_ = class_or_object if not isinstance(type_, type): type_ = type(class_or_object) assert isinstance(type_, type) return type_ def is_structured_config_frozen(obj: Any) -> bool: type_ = get_type_of(obj) if is_dataclass(type_): return is_dataclass_frozen(type_) if is_attr_class(type_): return is_attr_frozen(type_) return False def get_structured_config_init_field_names(obj: Any) -> List[str]: fields: Union[List["dataclasses.Field[Any]"], List["attr.Attribute[Any]"]] if is_dataclass(obj): fields = get_dataclass_fields(obj) elif is_attr_class(obj): fields = get_attr_class_fields(obj) else: raise ValueError(f"Unsupported type: {type(obj).__name__}") return [f.name for f in fields if f.init] def get_structured_config_data( obj: Any, allow_objects: Optional[bool] = None ) -> Dict[str, Any]: if is_dataclass(obj): return get_dataclass_data(obj, allow_objects=allow_objects) elif is_attr_class(obj): return get_attr_data(obj, allow_objects=allow_objects) else: raise ValueError(f"Unsupported type: {type(obj).__name__}") class ValueKind(Enum): VALUE = 0 MANDATORY_MISSING = 1 INTERPOLATION = 2 def _is_missing_value(value: Any) -> bool: from omegaconf import Node if isinstance(value, Node): value = value._value() return _is_missing_literal(value) def _is_missing_literal(value: Any) -> bool: # Uses literal '???' instead of the MISSING const for performance reasons. return isinstance(value, str) and value == "???" def _is_none( value: Any, resolve: bool = False, throw_on_resolution_failure: bool = True ) -> bool: from omegaconf import Node if not isinstance(value, Node): return value is None if resolve: value = value._maybe_dereference_node( throw_on_resolution_failure=throw_on_resolution_failure ) if not throw_on_resolution_failure and value is None: # Resolution failure: consider that it is *not* None. return False assert isinstance(value, Node) return value._is_none() def get_value_kind( value: Any, strict_interpolation_validation: bool = False ) -> ValueKind: """ Determine the kind of a value Examples: VALUE: "10", "20", True MANDATORY_MISSING: "???" INTERPOLATION: "${foo.bar}", "${foo.${bar}}", "${foo:bar}", "[${foo}, ${bar}]", "ftp://${host}/path", "${foo:${bar}, [true], {'baz': ${baz}}}" :param value: Input to classify. :param strict_interpolation_validation: If `True`, then when `value` is a string containing "${", it is parsed to validate the interpolation syntax. If `False`, this parsing step is skipped: this is more efficient, but will not detect errors. """ if _is_missing_value(value): return ValueKind.MANDATORY_MISSING if _is_interpolation(value, strict_interpolation_validation): return ValueKind.INTERPOLATION return ValueKind.VALUE def _is_interpolation(v: Any, strict_interpolation_validation: bool = False) -> bool: from omegaconf import Node if isinstance(v, Node): v = v._value() if isinstance(v, str) and _is_interpolation_string( v, strict_interpolation_validation ): return True return False def _is_interpolation_string(value: str, strict_interpolation_validation: bool) -> bool: # We identify potential interpolations by the presence of "${" in the string. # Note that escaped interpolations (ex: "esc: \${bar}") are identified as # interpolations: this is intended, since they must be processed as interpolations # for the string to be properly un-escaped. # Keep in mind that invalid interpolations will only be detected when # `strict_interpolation_validation` is True. if "${" in value: if strict_interpolation_validation: # First try the cheap regex matching that detects common interpolations. if SIMPLE_INTERPOLATION_PATTERN.match(value) is None: # If no match, do the more expensive grammar parsing to detect errors. parse(value) return True return False def _is_special(value: Any) -> bool: """Special values are None, MISSING, and interpolation.""" return _is_none(value) or get_value_kind(value) in ( ValueKind.MANDATORY_MISSING, ValueKind.INTERPOLATION, ) def is_float(st: str) -> bool: try: float(st) return True except ValueError: return False def is_int(st: str) -> bool: try: int(st) return True except ValueError: return False def is_primitive_list(obj: Any) -> bool: return isinstance(obj, (list, tuple)) def is_primitive_dict(obj: Any) -> bool: t = get_type_of(obj) return t is dict def is_dict_annotation(type_: Any) -> bool: if type_ in (dict, Dict): return True origin = getattr(type_, "__origin__", None) # type_dict is a bit hard to detect. # this support is tentative, if it eventually causes issues in other areas it may be dropped. if sys.version_info < (3, 7, 0): # pragma: no cover typed_dict = hasattr(type_, "__base__") and type_.__base__ == Dict return origin is Dict or type_ is Dict or typed_dict else: # pragma: no cover typed_dict = hasattr(type_, "__base__") and type_.__base__ == dict return origin is dict or typed_dict def is_list_annotation(type_: Any) -> bool: if type_ in (list, List): return True origin = getattr(type_, "__origin__", None) if sys.version_info < (3, 7, 0): return origin is List or type_ is List # pragma: no cover else: return origin is list # pragma: no cover def is_tuple_annotation(type_: Any) -> bool: if type_ in (tuple, Tuple): return True origin = getattr(type_, "__origin__", None) if sys.version_info < (3, 7, 0): return origin is Tuple or type_ is Tuple # pragma: no cover else: return origin is tuple # pragma: no cover def is_supported_union_annotation(obj: Any) -> bool: """Currently only primitive types are supported in Unions, e.g. Union[int, str]""" if not is_union_annotation(obj): return False args = obj.__args__ return all(is_primitive_type_annotation(arg) for arg in args) def is_dict_subclass(type_: Any) -> bool: return type_ is not None and isinstance(type_, type) and issubclass(type_, Dict) def is_dict(obj: Any) -> bool: return is_primitive_dict(obj) or is_dict_annotation(obj) or is_dict_subclass(obj) def is_primitive_container(obj: Any) -> bool: return is_primitive_list(obj) or is_primitive_dict(obj) def get_list_element_type(ref_type: Optional[Type[Any]]) -> Any: args = getattr(ref_type, "__args__", None) if ref_type is not List and args is not None and args[0]: element_type = args[0] else: element_type = Any return element_type def get_tuple_item_types(ref_type: Type[Any]) -> Tuple[Any, ...]: args = getattr(ref_type, "__args__", None) if args in (None, ()): args = (Any, ...) assert isinstance(args, tuple) return args def get_dict_key_value_types(ref_type: Any) -> Tuple[Any, Any]: args = getattr(ref_type, "__args__", None) if args is None: bases = getattr(ref_type, "__orig_bases__", None) if bases is not None and len(bases) > 0: args = getattr(bases[0], "__args__", None) key_type: Any element_type: Any if ref_type is None or ref_type == Dict: key_type = Any element_type = Any else: if args is not None: key_type = args[0] element_type = args[1] else: key_type = Any element_type = Any return key_type, element_type def is_valid_value_annotation(type_: Any) -> bool: _, type_ = _resolve_optional(type_) return ( type_ is Any or is_primitive_type_annotation(type_) or is_structured_config(type_) or is_container_annotation(type_) or is_supported_union_annotation(type_) ) def _valid_dict_key_annotation_type(type_: Any) -> bool: from omegaconf import DictKeyType return type_ is None or type_ is Any or issubclass(type_, DictKeyType.__args__) # type: ignore def is_primitive_type_annotation(type_: Any) -> bool: type_ = get_type_of(type_) return issubclass(type_, (Enum, pathlib.Path)) or type_ in BUILTIN_VALUE_TYPES def _get_value(value: Any) -> Any: from .base import Container, UnionNode from .nodes import ValueNode if isinstance(value, ValueNode): return value._value() elif isinstance(value, Container): boxed = value._value() if boxed is None or _is_missing_literal(boxed) or _is_interpolation(boxed): return boxed elif isinstance(value, UnionNode): boxed = value._value() if boxed is None or _is_missing_literal(boxed) or _is_interpolation(boxed): return boxed else: return _get_value(boxed) # pass through value of boxed node # return primitives and regular OmegaConf Containers as is return value def get_type_hint(obj: Any, key: Any = None) -> Optional[Type[Any]]: from omegaconf import Container, Node if isinstance(obj, Container): if key is not None: obj = obj._get_node(key) else: if key is not None: raise ValueError("Key must only be provided when obj is a container") if isinstance(obj, Node): ref_type = obj._metadata.ref_type if obj._is_optional() and ref_type is not Any: return Optional[ref_type] # type: ignore else: return ref_type else: return Any # type: ignore def _raise(ex: Exception, cause: Exception) -> None: # Set the environment variable OC_CAUSE=1 to get a stacktrace that includes the # causing exception. env_var = os.environ["OC_CAUSE"] if "OC_CAUSE" in os.environ else None debugging = sys.gettrace() is not None full_backtrace = (debugging and not env_var == "0") or (env_var == "1") if full_backtrace: ex.__cause__ = cause else: ex.__cause__ = None raise ex.with_traceback(sys.exc_info()[2]) # set env var OC_CAUSE=1 for full trace def format_and_raise( node: Any, key: Any, value: Any, msg: str, cause: Exception, type_override: Any = None, ) -> None: from omegaconf import OmegaConf from omegaconf.base import Node if isinstance(cause, AssertionError): raise if isinstance(cause, OmegaConfBaseException) and cause._initialized: ex = cause if type_override is not None: ex = type_override(str(cause)) ex.__dict__ = copy.deepcopy(cause.__dict__) _raise(ex, cause) object_type: Optional[Type[Any]] object_type_str: Optional[str] = None ref_type: Optional[Type[Any]] ref_type_str: Optional[str] child_node: Optional[Node] = None if node is None: full_key = key if key is not None else "" object_type = None ref_type = None ref_type_str = None else: if key is not None and not node._is_none(): child_node = node._get_node(key, validate_access=False) try: full_key = node._get_full_key(key=key) except Exception as exc: # Since we are handling an exception, raising a different one here would # be misleading. Instead, we display it in the key. full_key = f"" object_type = OmegaConf.get_type(node) object_type_str = type_str(object_type) ref_type = get_type_hint(node) ref_type_str = type_str(ref_type) msg = string.Template(msg).safe_substitute( REF_TYPE=ref_type_str, OBJECT_TYPE=object_type_str, KEY=key, FULL_KEY=full_key, VALUE=value, VALUE_TYPE=type_str(type(value), include_module_name=True), KEY_TYPE=f"{type(key).__name__}", ) if ref_type not in (None, Any): template = dedent( """\ $MSG full_key: $FULL_KEY reference_type=$REF_TYPE object_type=$OBJECT_TYPE""" ) else: template = dedent( """\ $MSG full_key: $FULL_KEY object_type=$OBJECT_TYPE""" ) s = string.Template(template=template) message = s.substitute( REF_TYPE=ref_type_str, OBJECT_TYPE=object_type_str, MSG=msg, FULL_KEY=full_key ) exception_type = type(cause) if type_override is None else type_override if exception_type == TypeError: exception_type = ConfigTypeError elif exception_type == IndexError: exception_type = ConfigIndexError ex = exception_type(f"{message}") if issubclass(exception_type, OmegaConfBaseException): ex._initialized = True ex.msg = message ex.parent_node = node ex.child_node = child_node ex.key = key ex.full_key = full_key ex.value = value ex.object_type = object_type ex.object_type_str = object_type_str ex.ref_type = ref_type ex.ref_type_str = ref_type_str _raise(ex, cause) def type_str(t: Any, include_module_name: bool = False) -> str: is_optional, t = _resolve_optional(t) if t is NoneType: return str(t.__name__) if t is Any: return "Any" if t is ...: return "..." if hasattr(t, "__name__"): name = str(t.__name__) elif getattr(t, "_name", None) is not None: # pragma: no cover name = str(t._name) elif getattr(t, "__origin__", None) is not None: # pragma: no cover name = type_str(t.__origin__) else: name = str(t) if name.startswith("typing."): # pragma: no cover name = name[len("typing.") :] args = getattr(t, "__args__", None) if args is not None: args = ", ".join( [type_str(t, include_module_name=include_module_name) for t in t.__args__] ) ret = f"{name}[{args}]" else: ret = name if include_module_name: if ( hasattr(t, "__module__") and t.__module__ != "builtins" and t.__module__ != "typing" and not t.__module__.startswith("omegaconf.") ): module_prefix = str(t.__module__) + "." else: module_prefix = "" ret = module_prefix + ret if is_optional: return f"Optional[{ret}]" else: return ret def _ensure_container(target: Any, flags: Optional[Dict[str, bool]] = None) -> Any: from omegaconf import OmegaConf if is_primitive_container(target): assert isinstance(target, (list, dict)) target = OmegaConf.create(target, flags=flags) elif is_structured_config(target): target = OmegaConf.structured(target, flags=flags) elif not OmegaConf.is_config(target): raise ValueError( "Invalid input. Supports one of " + "[dict,list,DictConfig,ListConfig,dataclass,dataclass instance,attr class,attr class instance]" ) return target def is_generic_list(type_: Any) -> bool: """ Checks if a type is a generic list, for example: list returns False typing.List returns False typing.List[T] returns True :param type_: variable type :return: bool """ return is_list_annotation(type_) and get_list_element_type(type_) is not None def is_generic_dict(type_: Any) -> bool: """ Checks if a type is a generic dict, for example: list returns False typing.List returns False typing.List[T] returns True :param type_: variable type :return: bool """ return is_dict_annotation(type_) and len(get_dict_key_value_types(type_)) > 0 def is_container_annotation(type_: Any) -> bool: return is_list_annotation(type_) or is_dict_annotation(type_) def split_key(key: str) -> List[str]: """ Split a full key path into its individual components. This is similar to `key.split(".")` but also works with the getitem syntax: "a.b" -> ["a", "b"] "a[b]" -> ["a", "b"] ".a.b[c].d" -> ["", "a", "b", "c", "d"] "[a].b" -> ["a", "b"] """ # Obtain the first part of the key (in docstring examples: a, a, .a, '') first = KEY_PATH_HEAD.match(key) assert first is not None first_stop = first.span()[1] # `tokens` will contain all elements composing the key. tokens = key[0:first_stop].split(".") # Optimization in case `key` has no other component: we are done. if first_stop == len(key): return tokens if key[first_stop] == "[" and not tokens[-1]: # This is a special case where the first key starts with brackets, e.g. # [a] or ..[a]. In that case there is an extra "" in `tokens` that we # need to get rid of: # [a] -> tokens = [""] but we would like [] # ..[a] -> tokens = ["", "", ""] but we would like ["", ""] tokens.pop() # Identify other key elements (in docstring examples: b, b, b/c/d, b) others = KEY_PATH_OTHER.findall(key[first_stop:]) # There are two groups in the `KEY_PATH_OTHER` regex: one for keys starting # with a dot (.b, .d) and one for keys starting with a bracket ([b], [c]). # Only one group can be non-empty. tokens += [dot_key if dot_key else bracket_key for dot_key, bracket_key in others] return tokens # Similar to Python 3.7+'s `contextlib.nullcontext` (which should be used instead, # once support for Python 3.6 is dropped). @contextmanager def nullcontext(enter_result: Any = None) -> Iterator[Any]: yield enter_result