[BUG]: typos sometimes detected and sometimes not on words in python files

[jerabaul29]

Issue Description

I use codebook-lsp with helix - first, thank you for an amazing tool and contribution to the community!

One thing I wonder: why are typos sometimes reported, and sometimes not? Is it somehow linked to the "kind" of the word in the syntax tree? See the examples below and the screenshot:

Operating System

Linux (Ubuntu/Debian)

Editor

Helix

Codebook Version

0.3.16

Configuration

[language-server.codebook]
command = "codebook-lsp"
args = ["serve"]

[[language]]
name = "python"
language-servers = ["ty", "ruff", "jedi", "pylsp", "codebook", "hx-lsp" ]

Steps to Reproduce

The dummy file I am using:

def some_func(a):
    "Some func is having typing issues"
    argwjh = a + " world"
    print(argwjh)


def iincorrect_name(a):
    print(a)


iincorrect_name(12)

iincorrect_var = "b"

Expected Behavior

Typos are not reported for:

iincorrect_name(12)

and

print(argwjh)

(see screenshot); I would expect these typos to be reported too.

Actual Behavior

The typos are sometimes reported, sometimes not, depending on the "role" of the mis-typed word.

Code Sample

Log Output

Additional Context

No response

[blopker]

Hey! Thanks for the kind words. This behavior is part of Codebooks philosophy, which could be better explained in the readme. Codebook aims to be unobtrusive by only marking issues the user controls. No sense in catching errors from external libraries and such. Part of this is to only mark issues for terms where they are defined, and not usage of those terms. Once those corrections are made, other, more language specific tooling can be used to fix usage.

As a nice side effect, this radically reduces the need to keep massive lists of language specific terms to skip, like all other spell checkers do.

Hope this helps!