is_dataclass check in transform() should only match instances not types

[RedBeard0531]

awesome-pattern-matching/apm/core.py

Lines 220 to 225 in a6dfd4a

	def transform(pattern, f):
	def rf(p):
	return transform(p, f)
	if is_dataclass(pattern):
	pattern = Dataclass(type(pattern), pattern.__dict__)

Normally you can just pass a type to a terse match, but currently it doesn't work if the type is a dataclass because it tries to match against the type as if it were an instance of the dataclass. https://docs.python.org/3/library/dataclasses.html#dataclasses.is_dataclass recommends using is_dataclass(obj) and not isinstance(obj, type) to only match instances.

Example showing the bug:

In [1]: from dataclasses import dataclass

In [2]: @dataclass
   ...: class Foo:
   ...:     x: int
   ...:

In [3]: from apm import *

In [4]: match(Foo(1), Foo, 1) # <--- should work but doesn't
---------------------------------------------------------------------------
MatchError                                Traceback (most recent call last)
<ipython-input-4-58e7e898a729> in <module>
----> 1 match(Foo(1), Foo, 1)

~/.local/lib/python3.9/site-packages/apm/match.py in match(value, pattern, multimatch, strict, captureall, *extra)
     83                 return invoke(acc[0], [])
     84             return acc[0]
---> 85         raise MatchError(value)
     86
     87     if isinstance(captureall, dict):

MatchError: Foo(x=1)

In [5]: match(Foo(1), InstanceOf(Foo), 1) # <--- workaround
Out[5]: 1

[scravy]

Thanks for the report and tracking it down so neatly. Fixed in e920779 and released as https://github.com/scravy/awesome-pattern-matching/releases/tag/0.21.2