435 lines
15 KiB
Python
435 lines
15 KiB
Python
|
from jedi.inference.cache import inference_state_method_cache
|
||
|
|
from jedi.inference.base_value import ValueSet, NO_VALUES, Value, \
|
||
|
|
iterator_to_value_set, LazyValueWrapper, ValueWrapper
|
||
|
|
from jedi.inference.compiled import builtin_from_name
|
||
|
|
from jedi.inference.value.klass import ClassFilter
|
||
|
|
from jedi.inference.value.klass import ClassMixin
|
||
|
|
from jedi.inference.utils import to_list
|
||
|
|
from jedi.inference.names import AbstractNameDefinition, ValueName
|
||
|
|
from jedi.inference.context import ClassContext
|
||
|
|
from jedi.inference.gradual.generics import TupleGenericManager
|
||
|
|
|
||
|
|
|
||
|
|
class _BoundTypeVarName(AbstractNameDefinition):
|
||
|
|
"""
|
||
|
|
This type var was bound to a certain type, e.g. int.
|
||
|
|
"""
|
||
|
|
def __init__(self, type_var, value_set):
|
||
|
|
self._type_var = type_var
|
||
|
|
self.parent_context = type_var.parent_context
|
||
|
|
self._value_set = value_set
|
||
|
|
|
||
|
|
def infer(self):
|
||
|
|
def iter_():
|
||
|
|
for value in self._value_set:
|
||
|
|
# Replace any with the constraints if they are there.
|
||
|
|
from jedi.inference.gradual.typing import AnyClass
|
||
|
|
if isinstance(value, AnyClass):
|
||
|
|
yield from self._type_var.constraints
|
||
|
|
else:
|
||
|
|
yield value
|
||
|
|
return ValueSet(iter_())
|
||
|
|
|
||
|
|
def py__name__(self):
|
||
|
|
return self._type_var.py__name__()
|
||
|
|
|
||
|
|
def __repr__(self):
|
||
|
|
return '<%s %s -> %s>' % (self.__class__.__name__, self.py__name__(), self._value_set)
|
||
|
|
|
||
|
|
|
||
|
|
class _TypeVarFilter:
|
||
|
|
"""
|
||
|
|
A filter for all given variables in a class.
|
||
|
|
|
||
|
|
A = TypeVar('A')
|
||
|
|
B = TypeVar('B')
|
||
|
|
class Foo(Mapping[A, B]):
|
||
|
|
...
|
||
|
|
|
||
|
|
In this example we would have two type vars given: A and B
|
||
|
|
"""
|
||
|
|
def __init__(self, generics, type_vars):
|
||
|
|
self._generics = generics
|
||
|
|
self._type_vars = type_vars
|
||
|
|
|
||
|
|
def get(self, name):
|
||
|
|
for i, type_var in enumerate(self._type_vars):
|
||
|
|
if type_var.py__name__() == name:
|
||
|
|
try:
|
||
|
|
return [_BoundTypeVarName(type_var, self._generics[i])]
|
||
|
|
except IndexError:
|
||
|
|
return [type_var.name]
|
||
|
|
return []
|
||
|
|
|
||
|
|
def values(self):
|
||
|
|
# The values are not relevant. If it's not searched exactly, the type
|
||
|
|
# vars are just global and should be looked up as that.
|
||
|
|
return []
|
||
|
|
|
||
|
|
|
||
|
|
class _AnnotatedClassContext(ClassContext):
|
||
|
|
def get_filters(self, *args, **kwargs):
|
||
|
|
filters = super().get_filters(
|
||
|
|
*args, **kwargs
|
||
|
|
)
|
||
|
|
yield from filters
|
||
|
|
|
||
|
|
# The type vars can only be looked up if it's a global search and
|
||
|
|
# not a direct lookup on the class.
|
||
|
|
yield self._value.get_type_var_filter()
|
||
|
|
|
||
|
|
|
||
|
|
class DefineGenericBaseClass(LazyValueWrapper):
|
||
|
|
def __init__(self, generics_manager):
|
||
|
|
self._generics_manager = generics_manager
|
||
|
|
|
||
|
|
def _create_instance_with_generics(self, generics_manager):
|
||
|
|
raise NotImplementedError
|
||
|
|
|
||
|
|
@inference_state_method_cache()
|
||
|
|
def get_generics(self):
|
||
|
|
return self._generics_manager.to_tuple()
|
||
|
|
|
||
|
|
def define_generics(self, type_var_dict):
|
||
|
|
from jedi.inference.gradual.type_var import TypeVar
|
||
|
|
changed = False
|
||
|
|
new_generics = []
|
||
|
|
for generic_set in self.get_generics():
|
||
|
|
values = NO_VALUES
|
||
|
|
for generic in generic_set:
|
||
|
|
if isinstance(generic, (DefineGenericBaseClass, TypeVar)):
|
||
|
|
result = generic.define_generics(type_var_dict)
|
||
|
|
values |= result
|
||
|
|
if result != ValueSet({generic}):
|
||
|
|
changed = True
|
||
|
|
else:
|
||
|
|
values |= ValueSet([generic])
|
||
|
|
new_generics.append(values)
|
||
|
|
|
||
|
|
if not changed:
|
||
|
|
# There might not be any type vars that change. In that case just
|
||
|
|
# return itself, because it does not make sense to potentially lose
|
||
|
|
# cached results.
|
||
|
|
return ValueSet([self])
|
||
|
|
|
||
|
|
return ValueSet([self._create_instance_with_generics(
|
||
|
|
TupleGenericManager(tuple(new_generics))
|
||
|
|
)])
|
||
|
|
|
||
|
|
def is_same_class(self, other):
|
||
|
|
if not isinstance(other, DefineGenericBaseClass):
|
||
|
|
return False
|
||
|
|
|
||
|
|
if self.tree_node != other.tree_node:
|
||
|
|
# TODO not sure if this is nice.
|
||
|
|
return False
|
||
|
|
given_params1 = self.get_generics()
|
||
|
|
given_params2 = other.get_generics()
|
||
|
|
|
||
|
|
if len(given_params1) != len(given_params2):
|
||
|
|
# If the amount of type vars doesn't match, the class doesn't
|
||
|
|
# match.
|
||
|
|
return False
|
||
|
|
|
||
|
|
# Now compare generics
|
||
|
|
return all(
|
||
|
|
any(
|
||
|
|
# TODO why is this ordering the correct one?
|
||
|
|
cls2.is_same_class(cls1)
|
||
|
|
# TODO I'm still not sure gather_annotation_classes is a good
|
||
|
|
# idea. They are essentially here to avoid comparing Tuple <=>
|
||
|
|
# tuple and instead compare tuple <=> tuple, but at the moment
|
||
|
|
# the whole `is_same_class` and `is_sub_class` matching is just
|
||
|
|
# not in the best shape.
|
||
|
|
for cls1 in class_set1.gather_annotation_classes()
|
||
|
|
for cls2 in class_set2.gather_annotation_classes()
|
||
|
|
) for class_set1, class_set2 in zip(given_params1, given_params2)
|
||
|
|
)
|
||
|
|
|
||
|
|
def get_signatures(self):
|
||
|
|
return []
|
||
|
|
|
||
|
|
def __repr__(self):
|
||
|
|
return '<%s: %s%s>' % (
|
||
|
|
self.__class__.__name__,
|
||
|
|
self._wrapped_value,
|
||
|
|
list(self.get_generics()),
|
||
|
|
)
|
||
|
|
|
||
|
|
|
||
|
|
class GenericClass(DefineGenericBaseClass, ClassMixin):
|
||
|
|
"""
|
||
|
|
A class that is defined with generics, might be something simple like:
|
||
|
|
|
||
|
|
class Foo(Generic[T]): ...
|
||
|
|
my_foo_int_cls = Foo[int]
|
||
|
|
"""
|
||
|
|
def __init__(self, class_value, generics_manager):
|
||
|
|
super().__init__(generics_manager)
|
||
|
|
self._class_value = class_value
|
||
|
|
|
||
|
|
def _get_wrapped_value(self):
|
||
|
|
return self._class_value
|
||
|
|
|
||
|
|
def get_type_hint(self, add_class_info=True):
|
||
|
|
n = self.py__name__()
|
||
|
|
# Not sure if this is the best way to do this, but all of these types
|
||
|
|
# are a bit special in that they have type aliases and other ways to
|
||
|
|
# become lower case. It's probably better to make them upper case,
|
||
|
|
# because that's what you can use in annotations.
|
||
|
|
n = dict(list="List", dict="Dict", set="Set", tuple="Tuple").get(n, n)
|
||
|
|
s = n + self._generics_manager.get_type_hint()
|
||
|
|
if add_class_info:
|
||
|
|
return 'Type[%s]' % s
|
||
|
|
return s
|
||
|
|
|
||
|
|
def get_type_var_filter(self):
|
||
|
|
return _TypeVarFilter(self.get_generics(), self.list_type_vars())
|
||
|
|
|
||
|
|
def py__call__(self, arguments):
|
||
|
|
instance, = super().py__call__(arguments)
|
||
|
|
return ValueSet([_GenericInstanceWrapper(instance)])
|
||
|
|
|
||
|
|
def _as_context(self):
|
||
|
|
return _AnnotatedClassContext(self)
|
||
|
|
|
||
|
|
@to_list
|
||
|
|
def py__bases__(self):
|
||
|
|
for base in self._wrapped_value.py__bases__():
|
||
|
|
yield _LazyGenericBaseClass(self, base, self._generics_manager)
|
||
|
|
|
||
|
|
def _create_instance_with_generics(self, generics_manager):
|
||
|
|
return GenericClass(self._class_value, generics_manager)
|
||
|
|
|
||
|
|
def is_sub_class_of(self, class_value):
|
||
|
|
if super().is_sub_class_of(class_value):
|
||
|
|
return True
|
||
|
|
return self._class_value.is_sub_class_of(class_value)
|
||
|
|
|
||
|
|
def with_generics(self, generics_tuple):
|
||
|
|
return self._class_value.with_generics(generics_tuple)
|
||
|
|
|
||
|
|
def infer_type_vars(self, value_set):
|
||
|
|
# Circular
|
||
|
|
from jedi.inference.gradual.annotation import merge_pairwise_generics, merge_type_var_dicts
|
||
|
|
|
||
|
|
annotation_name = self.py__name__()
|
||
|
|
type_var_dict = {}
|
||
|
|
if annotation_name == 'Iterable':
|
||
|
|
annotation_generics = self.get_generics()
|
||
|
|
if annotation_generics:
|
||
|
|
return annotation_generics[0].infer_type_vars(
|
||
|
|
value_set.merge_types_of_iterate(),
|
||
|
|
)
|
||
|
|
else:
|
||
|
|
# Note: we need to handle the MRO _in order_, so we need to extract
|
||
|
|
# the elements from the set first, then handle them, even if we put
|
||
|
|
# them back in a set afterwards.
|
||
|
|
for py_class in value_set:
|
||
|
|
if py_class.is_instance() and not py_class.is_compiled():
|
||
|
|
py_class = py_class.get_annotated_class_object()
|
||
|
|
else:
|
||
|
|
continue
|
||
|
|
|
||
|
|
if py_class.api_type != 'class':
|
||
|
|
# Functions & modules don't have an MRO and we're not
|
||
|
|
# expecting a Callable (those are handled separately within
|
||
|
|
# TypingClassValueWithIndex).
|
||
|
|
continue
|
||
|
|
|
||
|
|
for parent_class in py_class.py__mro__():
|
||
|
|
class_name = parent_class.py__name__()
|
||
|
|
if annotation_name == class_name:
|
||
|
|
merge_type_var_dicts(
|
||
|
|
type_var_dict,
|
||
|
|
merge_pairwise_generics(self, parent_class),
|
||
|
|
)
|
||
|
|
break
|
||
|
|
|
||
|
|
return type_var_dict
|
||
|
|
|
||
|
|
|
||
|
|
class _LazyGenericBaseClass:
|
||
|
|
def __init__(self, class_value, lazy_base_class, generics_manager):
|
||
|
|
self._class_value = class_value
|
||
|
|
self._lazy_base_class = lazy_base_class
|
||
|
|
self._generics_manager = generics_manager
|
||
|
|
|
||
|
|
@iterator_to_value_set
|
||
|
|
def infer(self):
|
||
|
|
for base in self._lazy_base_class.infer():
|
||
|
|
if isinstance(base, GenericClass):
|
||
|
|
# Here we have to recalculate the given types.
|
||
|
|
yield GenericClass.create_cached(
|
||
|
|
base.inference_state,
|
||
|
|
base._wrapped_value,
|
||
|
|
TupleGenericManager(tuple(self._remap_type_vars(base))),
|
||
|
|
)
|
||
|
|
else:
|
||
|
|
if base.is_class_mixin():
|
||
|
|
# This case basically allows classes like `class Foo(List)`
|
||
|
|
# to be used like `Foo[int]`. The generics are not
|
||
|
|
# necessary and can be used later.
|
||
|
|
yield GenericClass.create_cached(
|
||
|
|
base.inference_state,
|
||
|
|
base,
|
||
|
|
self._generics_manager,
|
||
|
|
)
|
||
|
|
else:
|
||
|
|
yield base
|
||
|
|
|
||
|
|
def _remap_type_vars(self, base):
|
||
|
|
from jedi.inference.gradual.type_var import TypeVar
|
||
|
|
filter = self._class_value.get_type_var_filter()
|
||
|
|
for type_var_set in base.get_generics():
|
||
|
|
new = NO_VALUES
|
||
|
|
for type_var in type_var_set:
|
||
|
|
if isinstance(type_var, TypeVar):
|
||
|
|
names = filter.get(type_var.py__name__())
|
||
|
|
new |= ValueSet.from_sets(
|
||
|
|
name.infer() for name in names
|
||
|
|
)
|
||
|
|
else:
|
||
|
|
# Mostly will be type vars, except if in some cases
|
||
|
|
# a concrete type will already be there. In that
|
||
|
|
# case just add it to the value set.
|
||
|
|
new |= ValueSet([type_var])
|
||
|
|
yield new
|
||
|
|
|
||
|
|
def __repr__(self):
|
||
|
|
return '<%s: %s>' % (self.__class__.__name__, self._lazy_base_class)
|
||
|
|
|
||
|
|
|
||
|
|
class _GenericInstanceWrapper(ValueWrapper):
|
||
|
|
def py__stop_iteration_returns(self):
|
||
|
|
for cls in self._wrapped_value.class_value.py__mro__():
|
||
|
|
if cls.py__name__() == 'Generator':
|
||
|
|
generics = cls.get_generics()
|
||
|
|
try:
|
||
|
|
return generics[2].execute_annotation()
|
||
|
|
except IndexError:
|
||
|
|
pass
|
||
|
|
elif cls.py__name__() == 'Iterator':
|
||
|
|
return ValueSet([builtin_from_name(self.inference_state, 'None')])
|
||
|
|
return self._wrapped_value.py__stop_iteration_returns()
|
||
|
|
|
||
|
|
def get_type_hint(self, add_class_info=True):
|
||
|
|
return self._wrapped_value.class_value.get_type_hint(add_class_info=False)
|
||
|
|
|
||
|
|
|
||
|
|
class _PseudoTreeNameClass(Value):
|
||
|
|
"""
|
||
|
|
In typeshed, some classes are defined like this:
|
||
|
|
|
||
|
|
Tuple: _SpecialForm = ...
|
||
|
|
|
||
|
|
Now this is not a real class, therefore we have to do some workarounds like
|
||
|
|
this class. Essentially this class makes it possible to goto that `Tuple`
|
||
|
|
name, without affecting anything else negatively.
|
||
|
|
"""
|
||
|
|
api_type = 'class'
|
||
|
|
|
||
|
|
def __init__(self, parent_context, tree_name):
|
||
|
|
super().__init__(
|
||
|
|
parent_context.inference_state,
|
||
|
|
parent_context
|
||
|
|
)
|
||
|
|
self._tree_name = tree_name
|
||
|
|
|
||
|
|
@property
|
||
|
|
def tree_node(self):
|
||
|
|
return self._tree_name
|
||
|
|
|
||
|
|
def get_filters(self, *args, **kwargs):
|
||
|
|
# TODO this is obviously wrong. Is it though?
|
||
|
|
class EmptyFilter(ClassFilter):
|
||
|
|
def __init__(self):
|
||
|
|
pass
|
||
|
|
|
||
|
|
def get(self, name, **kwargs):
|
||
|
|
return []
|
||
|
|
|
||
|
|
def values(self, **kwargs):
|
||
|
|
return []
|
||
|
|
|
||
|
|
yield EmptyFilter()
|
||
|
|
|
||
|
|
def py__class__(self):
|
||
|
|
# This might not be 100% correct, but it is good enough. The details of
|
||
|
|
# the typing library are not really an issue for Jedi.
|
||
|
|
return builtin_from_name(self.inference_state, 'type')
|
||
|
|
|
||
|
|
@property
|
||
|
|
def name(self):
|
||
|
|
return ValueName(self, self._tree_name)
|
||
|
|
|
||
|
|
def get_qualified_names(self):
|
||
|
|
return (self._tree_name.value,)
|
||
|
|
|
||
|
|
def __repr__(self):
|
||
|
|
return '%s(%s)' % (self.__class__.__name__, self._tree_name.value)
|
||
|
|
|
||
|
|
|
||
|
|
class BaseTypingValue(LazyValueWrapper):
|
||
|
|
def __init__(self, parent_context, tree_name):
|
||
|
|
self.inference_state = parent_context.inference_state
|
||
|
|
self.parent_context = parent_context
|
||
|
|
self._tree_name = tree_name
|
||
|
|
|
||
|
|
@property
|
||
|
|
def name(self):
|
||
|
|
return ValueName(self, self._tree_name)
|
||
|
|
|
||
|
|
def _get_wrapped_value(self):
|
||
|
|
return _PseudoTreeNameClass(self.parent_context, self._tree_name)
|
||
|
|
|
||
|
|
def get_signatures(self):
|
||
|
|
return self._wrapped_value.get_signatures()
|
||
|
|
|
||
|
|
def __repr__(self):
|
||
|
|
return '%s(%s)' % (self.__class__.__name__, self._tree_name.value)
|
||
|
|
|
||
|
|
|
||
|
|
class BaseTypingClassWithGenerics(DefineGenericBaseClass):
|
||
|
|
def __init__(self, parent_context, tree_name, generics_manager):
|
||
|
|
super().__init__(generics_manager)
|
||
|
|
self.inference_state = parent_context.inference_state
|
||
|
|
self.parent_context = parent_context
|
||
|
|
self._tree_name = tree_name
|
||
|
|
|
||
|
|
def _get_wrapped_value(self):
|
||
|
|
return _PseudoTreeNameClass(self.parent_context, self._tree_name)
|
||
|
|
|
||
|
|
def __repr__(self):
|
||
|
|
return '%s(%s%s)' % (self.__class__.__name__, self._tree_name.value,
|
||
|
|
self._generics_manager)
|
||
|
|
|
||
|
|
|
||
|
|
class BaseTypingInstance(LazyValueWrapper):
|
||
|
|
def __init__(self, parent_context, class_value, tree_name, generics_manager):
|
||
|
|
self.inference_state = class_value.inference_state
|
||
|
|
self.parent_context = parent_context
|
||
|
|
self._class_value = class_value
|
||
|
|
self._tree_name = tree_name
|
||
|
|
self._generics_manager = generics_manager
|
||
|
|
|
||
|
|
def py__class__(self):
|
||
|
|
return self._class_value
|
||
|
|
|
||
|
|
def get_annotated_class_object(self):
|
||
|
|
return self._class_value
|
||
|
|
|
||
|
|
def get_qualified_names(self):
|
||
|
|
return (self.py__name__(),)
|
||
|
|
|
||
|
|
@property
|
||
|
|
def name(self):
|
||
|
|
return ValueName(self, self._tree_name)
|
||
|
|
|
||
|
|
def _get_wrapped_value(self):
|
||
|
|
object_, = builtin_from_name(self.inference_state, 'object').execute_annotation()
|
||
|
|
return object_
|
||
|
|
|
||
|
|
def __repr__(self):
|
||
|
|
return '<%s: %s>' % (self.__class__.__name__, self._generics_manager)
|