Coder Social home page Coder Social logo

pytypes's People

Contributors

agronholm avatar cclauss avatar dbarnett avatar elcombato avatar kentzo avatar lubieowoce avatar mitar avatar qria avatar rover-debug avatar sjjessop avatar stewori avatar

Stargazers

 avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar

Watchers

 avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar

pytypes's Issues

Deploy to pypi?

This library looks quite useful; it would be awesome to have it pip-installable!

TypeChecker does not work

The following trivial script fails with TypeError: 'TypeChecker' object does not support indexing:

from pytypes import TypeChecker

with TypeChecker():
    pass

Python 3.5.5.
Surely some kind of basic testing would've been nice? This works in typeguard btw.

_issubclass and bound_typevars argument

import typing
from pytypes import type_util

T = typing.TypeVar('T', covariant=True)

class L(typing.List[T]):
    pass

C = typing.TypeVar('T', bound=L)

type_util._issubclass(L[float], C)  # False
type_util._issubclass(L[float], C, bound_typevars={})  # True

Why does bound_typevar change result from False to True?

Also, why code never checks if superclass is in bound_typevars? I think it just relays on catch-all except. I think such approach might hide some bugs?

Python >= 3.6.1 (maybe others too but not 3.6.0 or 3.5): Typecheck agent fails on re.compile()

The following code:

import re
from pytypes import TypeChecker

with TypeChecker():
    re.compile('a')

fails as follows:

$ python3 Test.py 
/usr/local/lib/python3.6/dist-packages/pytypes-1.0b5.post20-py3.6.egg/pytypes/type_util.py:2532: UserWarning: the system profiling hook has changed unexpectedly
  warn('the system profiling hook has changed unexpectedly')
Traceback (most recent call last):
  File "/usr/local/lib/python3.6/dist-packages/pytypes-1.0b5.post20-py3.6.egg/pytypes/typecomment_parser.py", line 52, in _get_typestrings
    srclines = inspect.getsourcelines(obj)[0]
  File "/usr/lib/python3.6/inspect.py", line 955, in getsourcelines
    lines, lnum = findsource(object)
  File "/usr/lib/python3.6/inspect.py", line 768, in findsource
    file = getsourcefile(object)
  File "/usr/lib/python3.6/inspect.py", line 684, in getsourcefile
    filename = getfile(object)
  File "/usr/lib/python3.6/inspect.py", line 666, in getfile
    'function, traceback, frame, or code object'.format(object))
TypeError: <slot wrapper '__and__' of 'int' objects> is not a module, class, method, function, traceback, frame, or code object

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
  File "Test.py", line 5, in <module>
    re.compile('a')
  File "/usr/lib/python3.6/re.py", line 233, in compile
    return _compile(pattern, flags)
  File "/usr/lib/python3.6/re.py", line 302, in _compile
    if not (flags & DEBUG):
  File "/usr/lib/python3.6/enum.py", line 803, in __and__
    def __and__(self, other):
  File "/usr/local/lib/python3.6/dist-packages/pytypes-1.0b5.post20-py3.6.egg/pytypes/type_util.py", line 2563, in __call__
    _check_caller_type(False, caller_level=self._caller_level_shift+1)
  File "/usr/local/lib/python3.6/dist-packages/pytypes-1.0b5.post20-py3.6.egg/pytypes/type_util.py", line 2416, in _check_caller_type
    cllable, clss = _find_typed_base_method(cllable, clss)
  File "/usr/local/lib/python3.6/dist-packages/pytypes-1.0b5.post20-py3.6.egg/pytypes/type_util.py", line 2131, in _find_typed_base_method
    if has_type_hints(util._actualfunc(fmeth)):
  File "/usr/local/lib/python3.6/dist-packages/pytypes-1.0b5.post20-py3.6.egg/pytypes/type_util.py", line 657, in has_type_hints
    return _has_type_hints(func0)
  File "/usr/local/lib/python3.6/dist-packages/pytypes-1.0b5.post20-py3.6.egg/pytypes/type_util.py", line 686, in _has_type_hints
    tpStr = _get_typestrings(func, False)
  File "/usr/local/lib/python3.6/dist-packages/pytypes-1.0b5.post20-py3.6.egg/pytypes/typecomment_parser.py", line 54, in _get_typestrings
    srclines = inspect.getsourcelines(getattr(obj.__class__, obj.__name__))[0]
AttributeError: type object 'wrapper_descriptor' has no attribute '__and__'

difference in is_subtype vs subclass for int and float

Is this intended behavior?

#Python 3.6.2 | packaged by conda-forge | (default, Jul 23 2017, 23:01:38)
#Type 'copyright', 'credits' or 'license' for more information
#IPython 6.2.1 -- An enhanced Interactive Python. Type '?' for help.

In [1]: import pytypes

In [2]: issubclass(int, float)
Out[2]: False

In [3]: pytypes.is_subtype(int, float)
Out[3]: True

check_argument_types broken on CPython 3.6.3

Just noticed that with CPyhton 3.6.3 tests related to check_argument_types started to fail.
I lastly tested with CPython 3.6.1, so not sure about CPython 3.6.2. Will test that one as soon as I find time...

Iterator and Generator annotations are not processed correctly

The following code:

from typing import Iterator

from pytypes import TypeChecker

def f() -> Iterator[str]:
    yield 'abc'

with TypeChecker():
    x = tuple(c for c in f())
    print("OK")

fails as follows:

$ python3 Test.py
Traceback (most recent call last):
  File "Test.py", line 9, in <module>
    x = tuple(c for c in f())
  File "Test.py", line 9, in <genexpr>
    x = tuple(c for c in f())
  File "Test.py", line 6, in f
    yield 'abc'
pytypes.exceptions.ReturnTypeError: 
  __main__.f
  returned incompatible type:
Expected: Iterator[str]
Received: str

Also a simple call to f(), not wrapped in generator expression and tuple, fails in a slightly different manner:

$ python3 Test.py
Exception ignored in: <generator object f at 0x7fe74e4b7db0>
Traceback (most recent call last):
  File "Test.py", line 5, in f
    def f() -> Iterator[str]:
  File "/usr/local/lib/python3.6/dist-packages/pytypes-1.0b5.post23-py3.6.egg/pytypes/type_util.py", line 2580, in __call__
    _check_caller_type(True, None, arg, caller_level=self._caller_level_shift+1)
  File "/usr/local/lib/python3.6/dist-packages/pytypes-1.0b5.post23-py3.6.egg/pytypes/type_util.py", line 2431, in _check_caller_type
    prop_getter, force_exception=True)
  File "/usr/local/lib/python3.6/dist-packages/pytypes-1.0b5.post23-py3.6.egg/pytypes/typechecker.py", line 718, in _checkfuncresult
    _raise_typecheck_error(msg, True, check_val, tpch, resSig, func)
  File "/usr/local/lib/python3.6/dist-packages/pytypes-1.0b5.post23-py3.6.egg/pytypes/type_util.py", line 2336, in _raise_typecheck_error
    raise pytypes.ReturnTypeError(msg)
pytypes.exceptions.ReturnTypeError: 
  __main__.f
  returned incompatible type:
Expected: Iterator[str]
Received: NoneType
OK

Similar stacks occur if Iterator[str] is replaced with Generator[str, None, None].

Tuple[(int, ...)] cannot be parsed on Python 2.7

Having a type comment like Tuple[(int, ...)] cannot be parsed on Python 2.7:

.../pytypes/typecomment_parser.py", line 225, in _funcsigtypesfromstring
    argTp = eval(argString, globals)
  File "<string>", line 1
    (Tuple[(int, ...)])

Gradual typing with Any?

Using pytypes-1.0b5 I get:

$ python
Python 3.6.5 (default, Mar 31 2018, 05:34:57) 
[GCC 7.3.0] on linux
Type "help", "copyright", "credits" or "license" for more information.
>>> import pytypes
>>> import typing
>>> T = typing.TypeVar('T')
>>> pytypes.is_subtype(T, typing.Any)
True
>>> pytypes.is_subtype(typing.Any, T)
False

Does pytypes support gradual typing? Is there a way to do verify an is-consistent-with relationship with its API?

Background: I'm trying to replace this method with a call to pytypes.is_subtype (using Python typing types):
https://github.com/apache/beam/blob/a843a08439eddcf10f140767761f8e8c1f88d715/sdks/python/apache_beam/typehints/typehints.py#L1112-L1120

Exception: Parameterized generics cannot be used with class or instance checks

Reproduction:

import typing
import pytypes
from pytypes import type_util

type_util._issubclass(typing.List[typing.Dict[str, str]], typing.List[dict])
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File ".../pytypes/type_util.py", line 1769, in _issubclass
    bound_typevars_readonly, follow_fwd_refs, _recursion_check)
  File ".../pytypes/type_util.py", line 1799, in _issubclass_2
    bound_typevars_readonly, follow_fwd_refs, _recursion_check)
  File ".../pytypes/type_util.py", line 1345, in _issubclass_Generic
    _recursion_check):
  File ".../pytypes/type_util.py", line 1769, in _issubclass
    bound_typevars_readonly, follow_fwd_refs, _recursion_check)
  File ".../pytypes/type_util.py", line 1796, in _issubclass_2
    _recursion_check)
  File ".../pytypes/type_util.py", line 1140, in _issubclass_Mapping_covariant
    return issubclass(subclass, superclass)
  File "/usr/lib/python3.6/typing.py", line 1148, in __subclasscheck__
    raise TypeError("Parameterized generics cannot be used with class "
TypeError: Parameterized generics cannot be used with class or instance checks

Using pytypes-1.0b5.

TypeError: isinstance expected 2 arguments, got 1

Running Usage example with decorator in readme gave following error for me.

Traceback (most recent call last):
  File "/tmp/tmp.py", line 31, in <module>
    pytypes.dump_cache()
NameError: name 'pytypes' is not defined
Error in atexit._run_exitfuncs:
Traceback (most recent call last):
  File "/Users/qria/.venvs/jupyter2/lib/python3.6/site-packages/pytypes/typelogger.py", line 316, in _dump_module
    assumed_glbls, implicit_globals, assumed_typevars)
  File "/Users/qria/.venvs/jupyter2/lib/python3.6/site-packages/pytypes/typelogger.py", line 755, in dump
    assumed_globals, implicit_globals, assumed_typevars)
  File "/Users/qria/.venvs/jupyter2/lib/python3.6/site-packages/pytypes/typelogger.py", line 680, in dump
    assumed_globals, implicit_globals))
  File "/Users/qria/.venvs/jupyter2/lib/python3.6/site-packages/pytypes/typelogger.py", line 653, in _stub_src_annotations
    assumed_globals, implicit_globals), ' ...']
  File "/Users/qria/.venvs/jupyter2/lib/python3.6/site-packages/pytypes/typelogger.py", line 625, in _declaration
    if annotated else self._signature(), ':'))
  File "/Users/qria/.venvs/jupyter2/lib/python3.6/site-packages/pytypes/typelogger.py", line 597, in _annotated_signature
    tp_lst = _prepare_arg_types_list(combine_argtype(self.arg_type_observations),
  File "/Users/qria/.venvs/jupyter2/lib/python3.6/site-packages/pytypes/typelogger.py", line 149, in combine_argtype
    assert isinstance(is_Tuple(observations[0]))
TypeError: isinstance expected 2 arguments, got 1

I don't know much about this codebase but maybe you've intended assert is_Tuple(observations[0])?

versions:

pytypes==1.0b4

README.rst contains utf-8 characters

So this means package cannot be installed on systems which do not have current local set to utf-8. In Python 3, you get an error like:

UnicodeDecodeError: 'ascii' codec can't decode byte 0xe2 in position 8634: ordinal not in range(128)

One option would be to do open(readme_path, encoding='utf8').read(), but this would not work on Python 2, where you will have to use io.open.

Other is that you remove all unicode characters from the README.

error with @autoclass concerning default arguments' type

I would like to provide an example showing how pytypes can be used with autoclass.
However the following code fails with a strange error:

from autoclass import autoclass, Boolean
from pytypes import typechecked
from numbers import Real, Integral
from typing import Optional

@typechecked
@autoclass
class HouseConfiguration(object):
    def __init__(self,
                 name: str,
                 surface: Real,
                 nb_floors: Optional[Integral] = 1,
                 with_windows: Boolean = False):
        pass

    # -- overriden setter for surface for custom validation
    @setter_override
    def surface(self, surface):
        assert surface > 0
        self._surface = surface

t = HouseConfiguration('test', 12, 2)  # error

The error received is :

Expected: Tuple[str, Real, Union[Integral, NoneType], Boolean]
Received: Tuple[str, int, int, int]

While this works:

t = HouseConfiguration('test', 12, nb_floors=2)

So it seems that this does not have anything to do with autoclass: somehow positional arguments / default values in the constructor signature are not handled properly.

Surprising issubclass result

Why the following returns false?

import typing
from pytypes import type_util

type_util._issubclass(typing.List[float], typing.List)

Isn't this the same as type_util._issubclass(typing.List[float], typing.List[typing.Any])? And then float is an instance of typing.Any? But this is false as well:

type_util._issubclass(typing.List[float], typing.List[typing.Any])

pytypes does not preserve the methods' signatures

from pytypes import typechecked
from inspect import signature

class Foo:
    def __init__(self, foo: str = 'hello'):
        self.foo = foo

s = signature(Foo.__init__)
print(s.parameters)
# > OrderedDict([('self', <Parameter "self">), ('foo', <Parameter "foo:str='hello'">)])
# parameter names, type hints and default values are preserved

Whereas

@typechecked
class Foo:
    def __init__(self, foo: str):
        self.foo = foo

s = signature(Foo.__init__)
print(s.parameters)
# > OrderedDict([('args', <Parameter "*args">), ('kw', <Parameter "**kw">)])
# everything has been removed ! 

This is annoying as it does not allow other libraries to automatically discover what is required to build an object. For example with parsyfiles.

The solution is to use a signature-preserving decorator library in pytypes, I personally use decorator but you can also use wrapt I think.

pytypes.always_check_parent_types is inconvenient for __init__. Add special treatment(?)

from pytypes import typechecked

@typechecked
class Class1():
    def __init__(self, arg1: str) -> None:
        pass


class Class2(Class1):
    def __init__(self, arg1: str, arg2: str) -> None:
        Class1.__init__(self, arg1)


Class2("arg1", "arg2")

Calling Class1.__init__(self, arg1) causes the following exception:

pytypes.exceptions.InputTypeError:
  __main__.Class2.__init__
  called with incompatible types:
Expected: Tuple[str, str]
Received: Tuple[str]

more tests

There is a lot of code in pytypes.type_utils. It would be great to see some tests to help understand it better, along with some docstrings (and type annotations!) on important functions. I was really surprised to see that there are no tests for pytypes.is_subtype .

Fault in typecheck agent on checking return type in presence of self or classmethod's cls

The following trivial code:

from configparser import ConfigParser
from pytypes import TypeChecker

with TypeChecker():
    ConfigParser()

fails as follows:

$ python3 Test.py
/usr/local/lib/python3.6/dist-packages/pytypes/type_util.py:2399: UserWarning: the system profiling hook has changed unexpectedly
  warn('the system profiling hook has changed unexpectedly')
Traceback (most recent call last):
  File "Test.py", line 5, in <module>
    ConfigParser()
  File "/usr/lib/python3.6/configparser.py", line 612, in __init__
    self._proxies[default_section] = SectionProxy(self, default_section)
  File "/usr/lib/python3.6/configparser.py", line 1223, in __init__
    for conv in parser.converters:
  File "/usr/lib/python3.6/configparser.py", line 1181, in converters
    return self._converters
  File "/usr/local/lib/python3.6/dist-packages/pytypes/type_util.py", line 2446, in __call__
    _check_caller_type(True, None, arg, caller_level=self._caller_level_shift+1)
  File "/usr/local/lib/python3.6/dist-packages/pytypes/type_util.py", line 2247, in _check_caller_type
    orig_clss = call_args[0].__orig_class__
  File "/usr/lib/python3.6/configparser.py", line 1306, in __getitem__
    return self._data[key]
KeyError: 0

An issue with forward declarations and recursive type

Example:

import typing

from pytypes import type_util

Container = typing.Union[
    typing.List['Data'],
]

Data = typing.Union[
    Container,
    str, bytes, bool, float, int, dict,
]

type_util._issubclass(typing.List[float], Container)
Traceback (most recent call last):
  File "<redacted>/lib/python3.6/site-packages/pytypes/type_util.py", line 1387, in _issubclass_2
    return issubclass(subclass, superclass)
TypeError: Forward references cannot be used with issubclass().

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
  File "test.py", line 14, in <module>
    type_util._issubclass(typing.List[float], Container)
TypeError: Invalid type declaration: float, _ForwardRef('Data')

List[int] is incompatible with List[Union[float]]

The following code:

from typing import cast, List

from pytypes import TypeChecker

def f() -> List[float]:
    return cast(List[float], [2])

with TypeChecker():
    f()
    print("OK")

fails as follows:

$ python3 Test.py
Traceback (most recent call last):
  File "Test.py", line 9, in <module>
    f()
  File "Test.py", line 6, in f
    return cast(List[float], [2])
pytypes.exceptions.ReturnTypeError: 
  __main__.f
  returned incompatible type:
Expected: List[float]
Received: List[int]

Normalize Union output of type_str

Union types are equivalent (but not identical) when they differ in the order of arguments. Currently, type_str prints out Union type with arguments in the same order as it is stored in the type. But I think it might be better (or at least optional with argument) to output it in a canonical and reproducible way (probably just sort it). This would allow one to output descriptions of strings without being surprised by random replacements of types with equivalent types. See here for more information: python/typing#559

So, my suggestion would be to add to type_str an argument canonical which would always output description with everything in the same order, no matter if you are outputting for equivalent but not identical types.

Error when checking a tuple against a typing.List

>>> from pytypes import type_util
>>> import typing
>>> type_util._isinstance((), typing.List)
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
TypeError: issubclass() arg 2 must be a class or tuple of classes
>>> type_util._isinstance((), typing.List[typing.Any])

is_of_type([], List[int]) throws an exception in 3.7

Python 3.7.0 (v3.7.0:1bf9cc5093, Jun 27 2018, 04:59:51) [MSC v.1914 64 bit (AMD64)] on win32
Type "help", "copyright", "credits" or "license" for more information.
>>> import typing
>>> import pytypes
>>> pytypes.is_of_type([], typing.List[int])
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "D:\.virtualenvs\pytypes\lib\site-packages\pytypes\type_util.py", line 1904, in _isinstance
    bound_typevars_readonly, follow_fwd_refs, _recursion_check)
  File "D:\.virtualenvs\pytypes\lib\site-packages\pytypes\type_util.py", line 1769, in _issubclass
    bound_typevars_readonly, follow_fwd_refs, _recursion_check)
  File "D:\.virtualenvs\pytypes\lib\site-packages\pytypes\type_util.py", line 1799, in _issubclass_2
    bound_typevars_readonly, follow_fwd_refs, _recursion_check)
  File "D:\.virtualenvs\pytypes\lib\site-packages\pytypes\type_util.py", line 1310, in _issubclass_Generic
    bound_typevars_readonly, follow_fwd_refs, _recursion_check):
  File "D:\.virtualenvs\pytypes\lib\site-packages\pytypes\type_util.py", line 1769, in _issubclass
    bound_typevars_readonly, follow_fwd_refs, _recursion_check)
  File "D:\.virtualenvs\pytypes\lib\site-packages\pytypes\type_util.py", line 1799, in _issubclass_2
    bound_typevars_readonly, follow_fwd_refs, _recursion_check)
  File "D:\.virtualenvs\pytypes\lib\site-packages\pytypes\type_util.py", line 1293, in _issubclass_Generic
    if subclass.__origin__ is None:
AttributeError: type object 'Empty' has no attribute '__origin__'

This seems to be related to #32. I tested the above in python 3.6.5 and it worked fine, so it's only an issue in python 3.7.

Error when running is_of_type on defaultdict

>>> import typing
>>> import collections
>>> import pytypes
>>> pytypes.is_of_type(collections.defaultdict(list), typing.Dict[str, typing.List])
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
TypeError: Parameterized generics cannot be used with class or instance checks

I would hope to get True instead.

Empty dict is incompatible with Mapping

The following code:

from typing import List, Mapping

from pytypes import TypeChecker

def f() -> List[Mapping[int, str]]:
    return [{}]

with TypeChecker():
    f()
    print("OK")

fails as follows:

$ python3 Test.py
Traceback (most recent call last):
  File "Test.py", line 9, in <module>
    f()
  File "Test.py", line 6, in f
    return [{}]
pytypes.exceptions.ReturnTypeError: 
  __main__.f
  returned incompatible type:
Expected: List[Mapping[int, str]]
Received: List[Empty[Dict]]

Checking generators throws an error

from pytypes import type_util
value = (i for i in range(10))
type_util._isinstance(value, int)

I would just assume it would be False, but it is:

TypeError: <code object <genexpr> at 0x7f5bc05418a0, file "test.py", line 2> is not a module, class, method, or function.

This one is even trickier:

from pytypes import type_util

class Foo:
    def bar(self):
        value = (i for i in range(10))
        type_util._isinstance(value, int)

Foo().bar()

get_Generic_parameters should return Any if no parmeters

If I understand correctly, the following class Bar has all parameters set to Any:

from typing import *

A = TypeVar('A')

class Foo(Generic[A]):
    pass

class Bar(Foo):
    pass

But pytypes.type_util.get_Generic_parameters(Bar, Foo) fails with:

TypeError: Bar has no proper parameters defined by Foo.

But it should return:

(typing.Any,)

From documentation:

Using a generic class without specifying type parameters assumes Any for each position. In the following example, MyIterable is not generic but implicitly inherits from Iterable[Any]:

from typing import Iterable

class MyIterable(Iterable): # Same as Iterable[Any]

get_Generic_parameters on Python 3.7

Per your comment ilevkivskyi/typing_inspect#35 (comment)

Using the latest master of pytypes (1449eff)

import typing as tp

import pytypes


T = tp.TypeVar("T")
T1 = tp.TypeVar("T1")
T2 = tp.TypeVar("T2")


class D0(tp.Dict[T1, T2]):
    def show_types(self) -> None:
        gt = pytypes.get_Generic_type(self)
        print(gt, pytypes.get_Generic_parameters(gt, D0))


class D1(D0[int, T]):
    pass


class D2(D1[str]):
    pass


D0[int, str]().show_types()
D1[str]().show_types()
D2().show_types()

works as expected on Python 3.6, producing:

__main__.D0[int, str] (<class 'int'>, <class 'str'>)
__main__.D1[str] (<class 'int'>, <class 'str'>)
__main__.D2 (<class 'int'>, <class 'str'>)

but crashes on Python 3.7:

__main__.D0[int, str] (<class 'int'>, <class 'str'>)
Traceback (most recent call last):
  File "pytypes10.py", line 26, in <module>
    D1[str]().show_types()
  File "pytypes10.py", line 14, in show_types
    print(gt, pytypes.get_Generic_parameters(gt, D0))
  File "venv-pytypes/lib/python3.7/site-packages/pytypes/type_util.py", line 327, in get_Generic_parameters
    res = _select_Generic_superclass_parameters(tp, generic_supertype)
  File "venv-pytypes/lib/python3.7/site-packages/pytypes/type_util.py", line 1319, in _select_Generic_superclass_parameters
    prms = _find_Generic_super_origin(subclass, superclass_origin)
  File "venv-pytypes/lib/python3.7/site-packages/pytypes/type_util.py", line 1256, in _find_Generic_super_origin
    if not bs.__origin__ is None:
AttributeError: type object 'D0' has no attribute '__origin__'

Maybe deserving a separate bug:

Changing

-class D1(D0[int, T]):
+class D1(D0[int, T1]):
    pass

causes the following erroneous output (running on Python 3.6):

__main__.D0[int, str] (<class 'int'>, <class 'str'>)
__main__.D1[str] (<class 'str'>, <class 'str'>)
__main__.D2 (<class 'str'>, <class 'str'>)

TypeVars should definitely be independent between classes (verified using reveal_type and mypy).

OrderedDict is incompatible with Mapping

The following code:

from collections import OrderedDict
from typing import Mapping

from pytypes import TypeChecker

def f() -> Mapping[int, str]:
    return OrderedDict()

with TypeChecker():
    f()

fails as follows:

$ python3 Test.py 
Traceback (most recent call last):
  File "Test.py", line 11, in <module>
    f()
  File "Test.py", line 8, in f
    return OrderedDict()
pytypes.exceptions.ReturnTypeError: 
  __main__.f
  returned incompatible type:
Expected: Mapping[int, str]
Received: OrderedDict

Python 3.7 support

Compatibility with the new upcoming typing module is not yet established. I'm working on this as far as time allows, help welcome.
928fd62 was the first milestone of this process, enabling the test suite to complete without a crash (not speaking of failing tests). Now, still most tests are failing. Work to do...

Typecheck agent: Incorrect reaction on `None` annotation of constructor return type

The following code:

from pytypes import TypeChecker

class A:
    def __init__(self) -> None:
        pass

with TypeChecker():
    A()

fails as follows:

$ python3 Test.py 
Traceback (most recent call last):
  File "Test.py", line 8, in <module>
    A()
  File "Test.py", line 5, in __init__
    pass
pytypes.exceptions.ReturnTypeError: 
  __main__.A.__init__
  returned incompatible type:
Expected: NoneType
Received: Tuple[A]

If None annotation to A.__init_() is removed, the fail disappears.

How to check if a type is a combination of some other types

In one part of my code, where I am deciding how to convert a value to JSON, I do:

any(is_subclass(structural_type, typ) for typ in (str, int, float, bool))

What I realized is that I would be OK with all possible Union combinations of these as well. So if structural_type is Union[str, int] that should also pass this check. Do you have a suggestion how to do so, or a helper function which would already do this? So I wonder about complicated case like Union[str, Union[int, bool]] or something like that.

List[function] is incompatible with List[Callable]

The following code:

from typing import Callable, List

from pytypes import TypeChecker

def n() -> None:
    pass

def f() -> List[Callable[[], None]]:
    return [n]

with TypeChecker():
    f()
    print("OK")

fails as follows:

$ python3 Test.py
Traceback (most recent call last):
  File "Test.py", line 12, in <module>
    f()
  File "Test.py", line 9, in f
    return [n]
pytypes.exceptions.ReturnTypeError: 
  __main__.f
  returned incompatible type:
Expected: List[Callable[[], NoneType]]
Received: List[function]

set() is an instance of typing.Sized

>>> import typing
>>> from pytypes import type_util
>>> type_util._isinstance(set(), typing.Sized)
False
>>> isinstance(set(), typing.Sized)
True

The first False is surprising. I would expect it to be True like isinstance is.

Frozenset seems to not be equal to set?

>>> import typing
>>> from pytypes import type_util
>>> type_util._isinstance(frozenset({1, 2, 'a', None, 'b'}), typing.AbstractSet[typing.Union[str, int, None]])
False
>>> type_util._isinstance({1, 2, 'a', None, 'b'}, typing.AbstractSet[typing.Union[str, int, type(None)]])
True
>>> issubclass(frozenset, typing.AbstractSet)
True

Empty tuple is incompatible with Tuple[Any, ...]

The following code:

from typing import Any, Tuple

from pytypes import TypeChecker

def f() -> Tuple[Any, ...]:
    return ()

with TypeChecker():
    f()
    print("OK")

fails as follows:

$ python3 Test.py
Traceback (most recent call last):
  File "Test.py", line 9, in <module>
    f()
  File "Test.py", line 6, in f
    return ()
pytypes.exceptions.ReturnTypeError: 
  __main__.f
  returned incompatible type:
Expected: Tuple[Any]
Received: Tuple[]

stub file load works only across single file

Problem presentation

The way the stub file implementation is done it will only work when all imports can be resolved using the Python files. A simple example when this is not the case is type aliases. For example imagine:

# util.pyi
from typing import Union

MagicParam = Union[int, float]
# magic.pyi
from .util import MagicParam

def a() -> MagicParam: ...

Loading magic.pyi will break with an import error of MagicParam as that is not defined inside the corresponding util.py. And given it's a type alias it should not be. As is used solely by the typing system.

The above example works with mypy, following the priority list described in PEP-561 resolution order.

Proposed solution

When loading stub files start with a clean import system, that first searches for pyi files, and if not present fallback to py. I have a prof of concept level of this here https://github.com/gaborbernat/prefer-stubs-python-import. If you think this request is reasonable I can make a PR for it. Thanks!

Support for non-parametrized generic types

from typing import Dict
from pytypes import typechecked

@typechecked
def foo(a: Dict):
    pass

foo({'name': 'test2', 'surface': 1})

Raises

   pytypes.exceptions.InputTypeError: 
     autoclass.tests.test_readme.foo
     called with incompatible types:
   Expected: Tuple[Dict]
   Received: Tuple[Dict[str, Union[str, int]]]

While it should not.
Note that it also makes PyCharm 2017.2.4 and 2017.3 crash when executed in the python terminal :

During handling of the above exception, another exception occurred:
Traceback (most recent call last):
  File "C:\Program Files\JetBrains\PyCharm Community Edition 2017.2.3\helpers\pydev\_pydev_bundle\pydev_console_utils.py", line 251, in add_exec
    more = self.do_add_exec(code_fragment)
  File "C:\Program Files\JetBrains\PyCharm Community Edition 2017.2.3\helpers\pydev\pydevconsole.py", line 123, in do_add_exec
    command.run()
  File "C:\Program Files\JetBrains\PyCharm Community Edition 2017.2.3\helpers\pydev\pydevconsole.py", line 82, in run
    self.more = self.interpreter.runsource(text, '<input>', symbol)
  File "C:\Miniconda3\envs\tools\lib\code.py", line 75, in runsource
    self.runcode(code)
  File "C:\Miniconda3\envs\tools\lib\code.py", line 95, in runcode
    self.showtraceback()
  File "C:\Miniconda3\envs\tools\lib\code.py", line 149, in showtraceback
    sys.excepthook(ei[0], ei[1], last_tb)
  File "C:\Miniconda3\envs\tools\lib\site-packages\pytypes\util.py", line 816, in _pytypes_excepthook
    traceback.print_exception(exctype, value, tb, _calc_traceback_limit(tb))
  File "C:\Miniconda3\envs\tools\lib\site-packages\pytypes\util.py", line 760, in _calc_traceback_limit
    if tb2.tb_next.tb_frame.f_code.co_filename.split(os.sep)[-2] == 'pytypes' and not \
IndexError: list index out of range

Is there a method to list all subclasses of a type, taking into account both parametrized, half-parametrized and not-parametrized generics ?

Ideally such a method would provide parameters to specify if the caller wishes to

  • only look at subclasses with the same status (subclasses with exactly the same response to is_generic() as the parent, and in case of a generic parent, with the same list of parameters)
  • or to extend to half- (if the parent is a non-parametrized generic, subclasses that are still generic and with at least one remaining parameter would be returned),
  • or to extend to entirely- parametrized subclasses

Issues type checking empty lists

>>> import typing
>>> from pytypes import type_util
>>> type_util._isinstance([], typing.Sequence)
True
>>> type_util._isinstance([], typing.Sequence[int])
False

False there is surprising. I understand why it happens: because internally isinstance is converted to issubclass and Sequence[Any] is not a subclass of Sequence[int].

Jython: ImportError: No module named pkg_resources

This is probably a Jython issue, but I'd like to have a fallback, since pytypes used to work with Jython apart from this.
As far as I understand, this is about some functionality of setuptools.
I didn't explicitly run setup.py on Jython, so something might be ill-configured.
However, I also didn't do such a thing on CPython, but there it runs out of the box.

How to fix it?

Found likely typo in pytypes.typechecker._typeinspect_func

Hi! while investigating a PR, I found this in checker_tp: (pytypes.typechecker._typeinspect_func:781)

if prop or prop_getter:
    slf = True
    util._warn_argname('property using non-idiomatic self argname',
            func0, slf, clsm)
    check_args = args_kw[1:] # omit self
 check_args = args_kw # <--- will always overwrite the `check_args` from the previous line!

I don't know enough about the logic involved to fix it, but I'm guessing it was intended to be this:

if prop or prop_getter:
    slf = True
    util._warn_argname('property using non-idiomatic self argname',
            func0, slf, clsm)
    check_args = args_kw[1:] # omit self
else:
    check_args = args_kw

(That also fixes some bugs I hit because of "non-idiomatic self" in @properties, but that's for another time)

Generator expression is not considered Iterable

The following code:

from typing import Iterable

from pytypes import TypeChecker

def f(a: Iterable[str]) -> None:
    pass

with TypeChecker():
    f((c for c in 'abc'))
    print("OK")

fails as follows:

$ python3 Test.py
Traceback (most recent call last):
  File "Test.py", line 9, in <module>
    f((c for c in 'abc'))
  File "Test.py", line 5, in f
    def f(a: Iterable[str]) -> None:
pytypes.exceptions.InputTypeError: 
  __main__.f
  called with incompatible types:
Expected: Tuple[Iterable[str]]
Received: Tuple[Generator]

typechecked methods of generics do not check typed iterators.

I think this might be an impossible issue to fix because Python iterators (and all other standard collections) can hold elements of multiple types. Feel free to close without much consideration.

If I have a class that extends generic and has a method that takes an iterator, I cannot type check that the iterator has all elements of the desired type. This arises in the use case of creating a typed list which supports extend.

from typing import TypeVar, Generic, Iterator
from pytypes import typechecked

T = TypeVar('T')

class TypList(Generic[T], list):

    @typechecked
    def append(self, obj: T) -> None:
        super().append(obj)

    @typechecked
    def extend(self, iterable: Iterator[T]) -> None:
        super().extend(iterable)

    @typechecked
    def insert(self, index: int, obj: T) -> None:
        super().insert(index, obj)

class IntList(TypList[int]):
    ...

il = IntList() 

il.append(1) # No error, as expected
il.append("a") # InputTypeError as expected

il.extend(iter([1, 2, 3])) # No error, as expected
il.extend(iter(["a", "b", "c"])) # No error, somewhat unexpected

print(il) # [1, 1, 2, 3, 'a', 'b', 'c']  <-- Ideally would only hold ints

In this case, a perfectly reasonable workaround is the check the contents of iterable manually, not using the decorator.

NoneType' object is not iterable in resolve_fw_decl

>>> import typing
>>> from pytypes import type_util
>>> T = typing.TypeVar('T')
>>> class Foo(typing.Generic[T]):
...   pass
... 
>>> type_util.resolve_fw_decl(Foo)
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
TypeError: 'NoneType' object is not iterable

Allow one to specify module to get_type_hints

Currently it is trying to guess the module, but this is not always possible. For example, it does not work for me if I call get_type_hints while the module is being defined and module does not yet exist.

A workaround for me is that currently I set im_class attribute on the function I want to get type hints for, but I think it might be cleaner if there would be an optional argument get_type_hints.

Recommend Projects

  • React photo React

    A declarative, efficient, and flexible JavaScript library for building user interfaces.

  • Vue.js photo Vue.js

    ๐Ÿ–– Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.

  • Typescript photo Typescript

    TypeScript is a superset of JavaScript that compiles to clean JavaScript output.

  • TensorFlow photo TensorFlow

    An Open Source Machine Learning Framework for Everyone

  • Django photo Django

    The Web framework for perfectionists with deadlines.

  • D3 photo D3

    Bring data to life with SVG, Canvas and HTML. ๐Ÿ“Š๐Ÿ“ˆ๐ŸŽ‰

Recommend Topics

  • javascript

    JavaScript (JS) is a lightweight interpreted programming language with first-class functions.

  • web

    Some thing interesting about web. New door for the world.

  • server

    A server is a program made to process requests and deliver data to clients.

  • Machine learning

    Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.

  • Game

    Some thing interesting about game, make everyone happy.

Recommend Org

  • Facebook photo Facebook

    We are working to build community through open source technology. NB: members must have two-factor auth.

  • Microsoft photo Microsoft

    Open source projects and samples from Microsoft.

  • Google photo Google

    Google โค๏ธ Open Source for everyone.

  • D3 photo D3

    Data-Driven Documents codes.