Linecount Report over counting

[sagarbadiyani]

Bug Report
The LineCountReporter (mypy.report.LineCountReporter) counts a function as typed even if one of the arguments is typed or it only has a return value (which I would refer as partially typed). Ideally, I would want a report of functions which are fully typed (and not partially) to gauge correctly how well-typed the codebase is.

To Reproduce
Create a python file test.py

from __future__ import annotations

def foo(a, b: str) -> str:
    return "bar"

Create an ini file .mypy.ini

[mypy]
files = test.py
linecount_report = .

Now run mypy --config-file .mypy.ini
And cat linecount.txt

Expected Behavior

I would expect the line count report to look like

      <don't care>       <don't care>      0      1 total

Actual Behavior

      <don't care>       <don't care>      1      1 total

Your Environment

• Mypy version used: 1.12.0
• Mypy command-line flags: --config-file
• Mypy configuration options from mypy.ini (and other config files):

[mypy]
files = test.py
linecount_report = .

• Python version used: 3.9.6

I believe this is happening because while visiting each function we are just checking for the existence of type attribute in its Function Definition

    def visit_func_def(self, defn: FuncDef) -> None:
        self.counts[defn.type is not None] += 1

I would expect this logic to be similar to the function def checker mypy.checker.TypeChecker.check_func_def; this method particularly mypy.checker.TypeChecker.check_for_missing_annotations which is actually a more elaborate check and goes over each argument to check the existence of its type

I am happy to raise a PR for this, but I need help with a couple of things

• Was this an intentional choice?
• If yes, can we consider adding a flag or reusing the disallow_incomplete_defs flag for determining if the function should be counted as typed or not in the report
• Even if no, this doesn't look like a backward compatible change, looking for thoughts around this
  • One of the ways to not make this a breaking change would be to use a new flag which is specific to linecount, in which case, of course, this is a feature request and not a bug

[sbadiyani-eightfold]

Requesting some traction on this issue
Tagging some of the active contributors I know: @JukkaL @hauntsaninja
Thank you 🙏🏻