configparser.SectionProxy.get* methods aren't correctly typed

[OddBloke]

On instantiation, configparser.SectionProxy takes as an argument the parser that it's a section of and uses setattr to populate itself with partially applied converters (i.e. methods that start with get) that the parser it is passed has.

Currently, this is stubbed as:

class SectionProxy(MutableMapping[str, str]):
    def __init__(self, parser: RawConfigParser, name: str) -> None: ...

    <snip>

    # SectionProxy can have arbitrary attributes when custon converters are used
    def __getattr__(self, key: str) -> Callable[..., Any]: ...

which means that ConfigParser()['section'].getboolean('key') type checks as Any, whereas it should in fact be bool.

[OddBloke]

I think we can cover the majority of use cases by simply copying the definitions down from RawConfigParser and removing the first argument (section, which is partially applied at runtime).

This won't help people who are subclassing RawConfigParser to add converters (as in https://github.com/OddBloke/jenkins-job-linter/blob/master/jenkins_job_linter/config.py#L26-L39). Those converters via SectionProxy will always type-check as returning Any; this isn't any worse than today.

[ambv]

Go for it.

[OddBloke]

Cool, looking now. The test case for this issue (thanks to @reddraggone9 in #1527 (comment)):

from configparser import ConfigParser

config = ConfigParser()
config.read_dict({'section': {'key': 'false'}})
assert config['section'].getboolean('key') is False

reveal_type(config['section'].getboolean('key'))
reveal_type(config['section'].getfloat('key'))
reveal_type(config['section'].getint('key'))

That currently outputs:

1543.py:7: error: Revealed type is 'Any'
1543.py:8: error: Revealed type is 'Any'
1543.py:9: error: Revealed type is 'Any'

but should output

1543.py:7: error: Revealed type is 'builtins.bool'
1543.py:8: error: Revealed type is 'builtins.float'
1543.py:9: error: Revealed type is 'builtins.int'

[OddBloke]

The commit above fixes this, but I committed it on top of #1527 so I'll wait for that to be merged to rebase and propose an MP.

[ambv]

#1527 is merged.