typing
typing copied to clipboard
python
Idea: add Char type to represent Text with size = 1
Due to the lack of character type in Python, a function that accept List[Text] will need to be extra careful when generalize the accepted argument type. Iterable[Text] and Sequence[Text] clearly doesn't work:
In [1]: import collections.abc
In [2]: isinstance("foo", collections.abc.Sequence)
Out[2]: True
In [3]: isinstance("foo", collections.abc.Iterable)
Out[3]: True
The closest thing you can do is to do something like Union[List[Text], Set[Text], Tuple[Text]] but it will not work with things like KeysView.
How about having a typing.Char to represent Text with length=1, and dedicate Text to a type with at least 2 characters?
Some examples to illustrate the usage:
def foo(xs: Iterable[Text]):
return '|'.join(xs)
# list(Text) should return List[Char]
foo("abcd") # error
foo(list("abcd")) # correct, List[Char] is a special case of List[Text]
def bar(xs: Iterable[Char]):
return '|'.join(xs)
bar("abcd") # correct, Text is a Iterable[Char]
bar(list("abcd")) # correct, List[Char] is a Iterable[Char]
bar(["ab", "cd"]) # error, List[Text] is not a Iterable[Char]
a: Char = 'a'
a[1] # error, Char cannot be indexed beyond index 0
a[0] # should be an error from type checking's PoV, Char should never be indexed
a[:] # should be an error from type checking's PoV, Char should never be copied as a slice
b: Text = 'b'
a + b # should be an error from type checking's PoV, Char and Text should be considered incompatible types
typing.cast(Text, a) + b # correct
c = a + a # correct, concatenate Char gets Text (but I'm not sure if this is reasonable though, what if we do (a + a) + a ?)
It is actually an interesting idea to have typing.Char as a string of length one. But for backwards compatibility we need to accept Char where Text is expected. I would recommend discussing this issue also on mypy tracker https://github.com/python/mypy
You can use phantom-types in this case: https://github.com/antonagestam/phantom-types/
from phantom.sized import PhantomSized
class Char(PhantomSized[str], str, len=1):
...
assert isinstance('a', Char) is True
assert isinstance('123', Char) is False
def works_with_char(c: Char): ...
