Incorrect Cognitive Complexity value when nesting for loops

Summary

I tried the cognitive_complexity lint on a basic code example. The result does not match up.

Input:

fn main() -> Result<()> {
    for _i in 0..10 { // +1
        for _k in 0..10 { // +2
            for _l in 0..10 { // +3
                for _m in 0..10 { // +4
                    for _n in 0..10 { // +5
                        for _o in 0..10 { // +6
                            print!("")
                        }
                    }
                }
            }
        }
    }

    Ok(())
}

Expected result: 21 Actual Result: 7

Additional Details To my understanding, the cognitive complexty of the main function is 21, as each for loop adds a value of 1, plus 1 for each level of nesting that it is already in.

However the warning I get says that it is 7:

warning: the function has a cognitive complexity of (7/1)
  --> src\main.rs:17:4
   |
17 | fn main() -> Result<()> {
   |    ^^^^
   |
note: the lint level is defined here
  --> src\main.rs:1:9
   |
1  | #![warn(clippy::cognitive_complexity)]
   |         ^^^^^^^^^^^^^^^^^^^^^^^^^^^^
   = help: you could split it up into multiple smaller functions
   = help: for further information visit https://rust-lang.github.io/rust-clippy/master/index.html#cognitive_complexity

Reproducer

I tried this code:

cargo clippy

Given this source code:

fn main() -> Result<()> {
    for _i in 0..10 { // +1
        for _k in 0..10 { // +2
            for _l in 0..10 { // +3
                for _m in 0..10 { // +4
                    for _n in 0..10 { // +5
                        for _o in 0..10 { // +6
                            print!("")
                        }
                    }
                }
            }
        }
    }

    Ok(())
}

I expected to see this happen: warning: the function has a cognitive complexity of (21/1)

Instead, this happened: warning: the function has a cognitive complexity of (7/1)

Version

rustc 1.71.0 (8ede3aae2 2023-07-12)
binary: rustc
commit-hash: 8ede3aae28fe6e4d52b38157d7bfe0d3bceef225
commit-date: 2023-07-12
host: x86_64-pc-windows-msvc
release: 1.71.0
LLVM version: 16.0.5

Additional Labels

No response

I found the following test (comments added by me). https://github.com/rust-lang/rust-clippy/blob/master/tests/ui/cognitive_complexity.rs#L91 https://github.com/rust-lang/rust-clippy/blob/master/tests/ui/cognitive_complexity.stderr#L10

fn kaboom() {
    let n = 0;
    'a: for i in 0..20 { // +1 (nesting level 0)
        'b: for j in i..20 { // +2 (nesting level 1)
            for k in j..20 { // +3 (nesting level 2)
                if k == 5 { // +4 (nesting level 3)
                    break 'b;
                }
                if j == 3 && k == 6 { // +5 (nesting level 3, +1 for introducting a boolean operation)
                    continue 'a;
                }
                if k == j { // +4 (nesting level 3)
                    continue;
                }
                println!("bake");
            }
        }
        println!("cake");
    }

    // Cognitive Complexity: 19
}

By counting I arrive at a value of 19, but the test suite asserts (and measures) a value of 7. Clearly there is some mismatch of definitions here. I go by the original specification https://www.sonarsource.com/docs/CognitiveComplexity.pdf

rust-lang / rust-clippy